chore(comments): drop epic, sprint and review-round ids from the code

The code cited its plan instead of saying why: `E28 review J4`, `(R6 S3b)`, `S1 justified…`, `deep-review round 5`. That plan lives in the private ops repo, so for every reader of this one the citation was noise, and a sentence whose subject was the plan step ("round 2 showed…") hid its lesson behind an id. Each one is gone and the "why" kept: a bare citation is deleted, a sentence carrying a lesson is reworded around it ("a later attempt showed…"). Measured (`scripts/audit_metrics.py`, product code): 652 lines in 270 files carried an epic code by the definition the next commit holds (507 / 241 by the old, two-digit-only one) -> 0. Across all code, tests, scripts and configs: ~1100 epic-code lines plus ~320 sprint/round ids. Behaviour unchanged, checked mechanically: every changed .py file has the same AST once docstrings are blanked; css/sh/toml/yml/json files are identical once comments are stripped; the only non-comment changes in TS/JS are test titles, one description string in a contract table, and five ESLint messages, whose text is all the gates snapshot diff holds. Not touched: Markdown (docs and the changelog keep their history), Alembic migrations, TV episode notation (`S2 · E5`), DESIGN.md ids (`C1`, `B4`, `S3h`), tracker ids (`SEC-16`, `DX-17`). One stale claim fixed on the way: test_hls_reaper's docstring said the player checkpoints "every 10 s whether or not it is playing", which plex/stream.py calls wrong; it now says what the code does.
2026-10-05 00:19:39 +02:00
parent c73d8be094
commit 4aaaaa7860
669 changed files with 1831 additions and 1904 deletions
@@ -46,7 +46,7 @@ describe("ActionMenu", () => {
    // A `title` on a disabled row reached nobody: a disabled button fires no pointer events and
    // roving focus skips it. The reason is visible text now, undimmed so it can be read. An enabled
    // row's state ("Downloaded" under "Download…") is the same line: a `title` shows to a resting
-    // mouse only, never to a key or a finger (round 2, F5).
+    // mouse only, never to a key or a finger.
    renderMenu([
      { id: "move", label: "Move to", onSelect: vi.fn(), disabled: true, note: "No other list" },
      { id: "dl", label: "Download…", onSelect: vi.fn(), note: "Downloaded" },
@@ -76,7 +76,7 @@ describe("ActionMenu", () => {

  // Close first, act on a later task: floating-ui hands the focus back to the trigger from a microtask
  // as the menu unmounts, and an action run in that same batch (a confirm dialog) had its focus taken
-  // back to the trigger behind the dialog's scrim — measured in Chrome, E11 S2. jsdom does not run
+  // back to the trigger behind the dialog's scrim — measured in Chrome. jsdom does not run
  // that focus race, so the ORDER is what is pinned: the menu is gone before the action runs.
  it("selecting an item closes the menu first, then fires its onSelect once", async () => {
    const onSelect = vi.fn();
@@ -182,7 +182,7 @@ describe("ActionMenu", () => {
  });
 });

-// PageUp/PageDown in a roving menu (E11, owner request) — the same page as the palette's and the
+// PageUp/PageDown in a roving menu (owner request) — the same page as the palette's and the
 // combobox's (lib/listPaging); here the keyboard's row holds real DOM focus (the `tabindex=0` rover).
 describe("ActionMenu · paging", () => {
  it("moves the roving focus a page at a time, over disabled rows, stopping at the ends", () => {
@@ -8,12 +8,12 @@ import { Pressable } from "./Pressable";
 import { SURFACE } from "./surfaces";

 // A reusable "⋯ more actions" overflow menu: a compact trigger that opens a list of one-shot actions.
-// Built on the R8 AnchoredPopover platform (arrows/typeahead/Escape/anchor-flip), same as ViewSwitcher
+// Built on the AnchoredPopover platform (arrows/typeahead/Escape/anchor-flip), same as ViewSwitcher
 // and LanguageSwitcher. Use it to keep occasional actions out of a crowded toolbar. The items are
 // fire-and-close (`onSelect`) — all but a `check: { kind: "many" }` row, which toggles in place; a
 // `disabled` item (e.g. a mutation already in flight) is inert.
 //
-// E10 S2 split the ITEM LIST out of the trigger (`MenuItemList`) so the ⋯ menu and the right-click
+// The ITEM LIST is split out of the trigger (`MenuItemList`) so the ⋯ menu and the right-click
 // context menu (`ContextMenu`) render the same rows from the same item type. That is the whole point
 // of the split: a context menu that drifts from the overflow menu beside it is two menus, and the
 // user has to learn both.
@@ -46,11 +46,11 @@ interface MenuItemBase {
  title?: string;
  /** Why a DISABLED row is off, shown as a line under its label. Not a `title`: a disabled button
   *  fires no pointer events (no tooltip) and roving focus skips it, so a reason there reached
-   *  nobody. An `aria-disabled` row was tried and reverted (E10 review rounds 1–2: floating-ui skips
+   *  nobody. An `aria-disabled` row was tried and reverted (floating-ui skips
   *  it identically, and the mouse got a highlight that then no-opped). Visible text is the one
   *  place a disabled row's reason can be: seen, and read in a screen reader's browse mode.
   *  On an ENABLED row it is the state of what the row opens ("Downloaded" under "Download…"),
-   *  for the same reason: a `title` shows to a resting mouse only (round 2, F5). */
+   *  for the same reason: a `title` shows to a resting mouse only. */
  note?: string;
  /** Draw a divider above this item. Ignored on the first item — a menu never opens with a rule. */
  separatorBefore?: boolean;
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// Which surfaces give the focus back to a card focused as a whole (deep-review E11 rounds 2 and 3). A
+// Which surfaces give the focus back to a card focused as a whole (found in code review). A
 // menu does, by hand — floating-ui's own return would land on the card's first link (see
 // ContextMenu.dom.test). A hover card never returned focus, and one closing after the reader clicked a
 // link in it must not pull the focus back to the card under it.
@@ -18,7 +18,7 @@ import {
 import { useFullscreenPortalHost } from "../../lib/fullscreen";
 import type { useAnchoredPopover } from "./anchoredPopoverHooks";

-// R8 popover platform. ONE anchored-popover primitive over @floating-ui/react, replacing the ~9
+// The popover platform. ONE anchored-popover primitive over @floating-ui/react, replacing the ~9
 // hand-rolled positioning engines (measured flip / width clamp / rAF scroll-follow / scroller-relative
 // visibility) that had each re-derived the same maths. Floating UI's middleware does all four:
 //   - flip()  → measured two-way flip (top/bottom) when a side runs out of room
@@ -28,9 +28,9 @@ import type { useAnchoredPopover } from "./anchoredPopoverHooks";
 //   - autoUpdate → follows the trigger on scroll/resize (the old rAF loop), coalesced by the library
 // Positioning is portalled through FloatingPortal (escapes ancestor stacking contexts — same job the
 // house <Overlay> did for TagsCell). Every popover is a node in the app's <FloatingTree> so Escape /
-// outside-press dispatch is topmost-only across nested layers (S3 folds the Modal + players in too).
+// outside-press dispatch is topmost-only across nested layers (the Modal and players included).
 //
-// S2 (this layer) adds the keyboard/ARIA model the roles promise, mirroring `ViewSwitcher` (the house's
+// This layer adds the keyboard/ARIA model the roles promise, mirroring `ViewSwitcher` (the house's
 // textbook menu) but via Floating UI's idioms rather than a second hand-rolled roving engine:
 //   - useListNavigation → roving focus over the registered items (↑/↓/Home/End, loop), menu/listbox only;
 //     PageUp/PageDown (a page, no loop) are ours, for every list surface incl. the combobox
@@ -45,9 +45,9 @@ import type { useAnchoredPopover } from "./anchoredPopoverHooks";
 //
 // The trigger takes its ref DESTRUCTURED: `const { setReference } = pop.refs;` then
 // `ref={setReference} {...pop.getReferenceProps()}`. Written `ref={pop.refs.setReference}`, the React
-// Compiler treats the whole `pop` as a ref and skips the component (E32 S3 — 21 call sites did).
+// Compiler treats the whole `pop` as a ref and skips the component (21 call sites once did).
 //
-// S? adds a "combobox" mode — the ONE right way to build a typeable combobox (a search/filter input
+// A "combobox" mode — the ONE right way to build a typeable combobox (a search/filter input
 // with a listbox of options below), so a hand-rolled floating-ui engine (the old MultiSelectChips /
 // PlexSearchBox pattern) is no longer needed. It differs from menu/listbox in exactly the focus model:
 //   - the trigger IS a text input, so open is driven by the caller (focus/typing), NOT useClick;
@@ -93,7 +93,7 @@ type ItemRole = "menuitem" | "menuitemradio" | "menuitemcheckbox" | "option";
 *  key. (Since the Tailwind v4 migration `bg-accent/NN` works — v4 does the opacity modifier via
 *  color-mix — so the recipe no longer needs a hand-written color-mix arbitrary value.) */
 // The roving-focus highlight. A hovered/roving row gets the SAME strong accent fill whether or not it
-// is selected — a checked row no longer reads as a weak ring-only afterthought (R10 weak-ring fix). The
+// is selected — a checked row no longer reads as a weak ring-only afterthought. The
 // selected state stays legible ON the fill because these menus mark "selected" with a CHILD element (a
 // checkmark / checkbox) that paints over the row tint; the selected row just gets a stronger ring
 // (ring-2, accent/80) than an unselected one (ring-1, accent/60) to keep the two apart.
@@ -242,7 +242,7 @@ interface PopoverSurfaceProps {
  passthrough?: boolean;
  /** The panel's role (a combobox's `listbox`) goes on an inner `<PopoverList>` instead of the panel
   *  itself, so the panel can also hold lines ABOUT the list — loading, nothing found, a failed load
-   *  with its Retry — outside it: a listbox may own options only (round 2, F4). */
+   *  with its Retry — outside it: a listbox may own options only. */
  listInside?: boolean;
 }

@@ -308,7 +308,7 @@ function PopoverSurfaceInner({
  // handled by the element-identity dependency. (Keying the portal to remount instead would rebuild
  // the subtree and lose all of that.) Focus does NOT survive on its own — `appendChild` on an
  // already-parented node is remove-then-insert, which blurs to <body>; the hook saves and restores
-  // it. This used to be a second copy of Modal's re-parent that lacked exactly that (round 5).
+  // it. This used to be a second copy of Modal's re-parent that lacked exactly that.
  const container = useFullscreenPortalHost();

  const { style: floatingPropStyle, ...floatingProps } = getFloatingProps() as {
@@ -336,7 +336,7 @@ function PopoverSurfaceInner({
  // but when that element is not a Tab stop itself, to its first TABBABLE descendant instead, even
  // when handed the element as a ref (its return runs through `getFirstTabbableElement` either way).
  // A card the keyboard is on (lib/listNav gives its root tabindex=-1) is exactly that: measured in
-  // Chrome (deep-review E11 round 2), j, Shift+F10, Escape left focus on the card's thumbnail link, the
+  // Chrome, j, Shift+F10, Escape left focus on the card's thumbnail link, the
  // ring off the card and the next Enter on a control. So for such an element the library's return is
  // off and this gives the focus back to it as it is — captured at open, before focus moves into the
  // panel, and only when the focus was still in a floating layer (or nowhere) as the panel closed:
@@ -345,7 +345,7 @@ function PopoverSurfaceInner({
  //
  // Only for a focus-managed surface, whose return this replaces: a hover card or a combobox never
  // returned focus, and a hover card closing after the reader clicked a link in it must not pull the
-  // focus back to the card under it (deep-review E11 round 3).
+  // focus back to the card under it.
  const [returnTo] = useState(() => {
    const active = document.activeElement;
    return managed &&
@@ -360,7 +360,7 @@ function PopoverSurfaceInner({
    return () => {
      // "Still in the layer" is read off the portal marker, not the panel: FloatingPortal renders the
      // panel a commit after this surface mounts, so the panel's ref is still empty when this effect
-      // runs (deep-review E11 round 3).
+      // runs.
      const active = document.activeElement;
      const stillOurs =
        !active || active === document.body || !!active.closest("[data-floating-ui-portal]");
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 review F7: while hidden the button only went transparent — a Tab still landed on the invisible
+// While hidden the button only went transparent — a Tab still landed on the invisible
 // control. And a keyboard press scrolled to the top and hid it with the focus still on it.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { act, cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -44,7 +44,7 @@ export default function BackToTop({ dep, threshold = 600 }: { dep?: unknown; thr
  }, [dep, threshold]);

  // A keyboard press hides the button it was on: the focus goes to the page it scrolled (its <main>
-  // is a focus rest, not a Tab stop), not left on an invisible control (E28 review F7).
+  // is a focus rest, not a Tab stop), not left on an invisible control.
  const toTop = (e: React.MouseEvent) => {
    const main = scrollerRef.current ?? document.querySelector("main");
    main?.scrollTo({ top: 0, behavior: "smooth" });
@@ -1,7 +1,7 @@
 import clsx from "clsx";
 import type { ReactNode } from "react";

-// A static label (a topic, a keyword, a genre, a status): flat and tonal, with NO edge (owner, E28 S3d).
+// A static label (a topic, a keyword, a genre, a status): flat and tonal, with NO edge (owner's call).
 // Every chip that does something has an edge, so "edged = pressable, flat = just says" holds at a
 // glance; a static tag painted like a chip is clicked and does nothing. The tint is the text colour
 // at 8%, so it reads on the dark schemes and on white alike; `onDark` is the always-dark player's. A
@@ -1,8 +1,8 @@
 // @vitest-environment jsdom
 //
-// S5: the shared <Button> primitive. These lock the load-bearing contract the migrated call sites rely
-// on: variant/size paint (the E28 B1 ladder and B3 layers), the loading state (disabled + spinner +
-// aria-busy), the pressed toggle, and the type="button" default that stops an action button from
+// The shared <Button> primitive. These lock the load-bearing contract the migrated call sites rely
+// on: variant/size paint (the DESIGN.md B1 ladder and B3 layers), the loading state (disabled +
+// spinner + aria-busy), the pressed toggle, and the type="button" default that stops an action button from
 // submitting a surrounding form. A glyph-only button is `IconButton` (its own test).
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { act, render, screen, cleanup, fireEvent } from "@testing-library/react";
@@ -10,7 +10,7 @@ import { Button } from "./Button";

 afterEach(cleanup);

-describe("Button primitive (S5)", () => {
+describe("Button primitive", () => {
  it("defaults to type=button so it never submits a form", () => {
    render(<Button>Go</Button>);
    expect(screen.getByRole("button", { name: "Go" }).getAttribute("type")).toBe("button");
@@ -85,7 +85,7 @@ describe("Button primitive (S5)", () => {
    expect(onClick).not.toHaveBeenCalled();
  });

-  // E11 S2, measured in Chrome: a focused button that becomes natively disabled drops the focus to
+  // Measured in Chrome: a focused button that becomes natively disabled drops the focus to
  // <body> and never gets it back, so every "Something else" press threw a keyboard user out of Home.
  // jsdom has no focus fixup, so what is pinned here is the cause — native `disabled` stays off.
  it("loading keeps the button natively enabled, so it keeps the focus", () => {
@@ -159,7 +159,7 @@ describe("Button primitive (S5)", () => {
  it("carries the coarse-pointer minimum target and the 24 floor, a link variant too", () => {
    // The tablet comfort contract: the minimum lives in the primitive, so no call site can opt
    // out of it by accident. A link variant was once excluded as "text in a sentence", but every
-    // one stands alone, and the E28 audit measured them at 16–20px.
+    // one stands alone, and an audit measured them at 16–20px.
    const { unmount } = render(<Button>Tap</Button>);
    expect(screen.getByRole("button", { name: "Tap" }).className).toContain("touch-target");
    unmount();
@@ -180,16 +180,16 @@ describe("Button primitive (S5)", () => {

 // --- the accessible-name contract, pinned at the TYPE level ------------------------------------
 // A button that renders no text must carry a name, and that rule is a union in Button.tsx rather
-// than a doc comment because the comment version let two nameless icon buttons ship (E10 review
-// round 1). But a type with no test is a type someone widens to silence a stubborn call site:
-// replacing `ButtonProps` with the permissive shape left tsc, eslint, knip and vitest all green
-// (round 2). Each `@ts-expect-error` below FAILS THE BUILD if the error it expects stops happening,
+// than a doc comment because the comment version let two nameless icon buttons ship. But a
+// type with no test is a type someone widens to silence a stubborn call site:
+// replacing `ButtonProps` with the permissive shape left tsc, eslint, knip and vitest all green.
+// Each `@ts-expect-error` below FAILS THE BUILD if the error it expects stops happening,
 // so weakening the union turns the whole typecheck lane red.
 //
 // Rendered, not just declared, so eslint sees them used and the file stays one test suite.
 describe("Button · disabled while it holds the focus", () => {
-  // A focused button that turns natively disabled loses the focus to <body> in Chrome (measured, E11
-  // S2) — and ~60 call sites disable the very button that was just pressed while its mutation runs.
+  // A focused button that turns natively disabled loses the focus to <body> in Chrome (measured)
+  // — and ~60 call sites disable the very button that was just pressed while its mutation runs.
  // jsdom does not move the focus, so the CONTRACT is what is pinned: while focused, the button is
  // unavailable by aria-disabled + a click guard, never by the native attribute.
  it("keeps the native attribute off while focused, and blocks activation", () => {
@@ -255,7 +255,7 @@ describe("Button accessible-name contract (type level)", () => {
      <>
        {/* @ts-expect-error an icon with no children renders no text and no name */}
        <Button icon={<span />} data-testid="bad-c" />
-        {/* @ts-expect-error the icon-only sizes moved to IconButton (E28 S3b) */}
+        {/* @ts-expect-error the icon-only sizes moved to IconButton */}
        <Button size="icon" label="gone" />
        <Button icon={<span />} label="by label" />
        <Button icon={<span />} aria-label="by aria-label" />
@@ -10,7 +10,7 @@ import { buttonClass, isLinkVariant, type ButtonSize, type ButtonVariant } from
 // holds. The app's other button families have their own primitives beside it: `IconButton` (a glyph
 // with no text), `Pressable` (a card, row or tile whose content is the control), `Chip` and
 // `FilterChipGroup` (filters), `SegmentedToggle`, `DragHandle`. A hand-rolled `<button>` is none of
-// them (E28 S3b).
+// them.
 //
 // No focus-visible here on purpose: @layer base in index.css already rings every <button> for keyboard
 // users. Adding one would double the ring.
@@ -93,7 +93,7 @@ describe("Carousel", () => {
  });
 });

-// E11 S4b: a rail on Home, Discover or an info page is a list the keyboard walks — but only when the
+// A rail on Home, Discover or an info page is a list the keyboard walks — but only when the
 // caller says its cards carry `navItemAttrs`. A carousel of plain content must not claim `j`.
 describe("Carousel · keyboard", () => {
  const cards = () =>
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3b: the chip family (docs/DESIGN.md "One primitive per button family"). One geometry for a
+// The chip family (docs/DESIGN.md "One primitive per button family"). One geometry for a
 // filter, an applied narrowing, a tag and a suggestion, where five files had five.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { act, cleanup, fireEvent, render, screen, waitFor } from "@testing-library/react";
@@ -42,7 +42,7 @@ describe("Chip", () => {
    expect(onClick).toHaveBeenCalledOnce();
  });

-  // Round 2 (F6): the feed's and Plex's active-filter chips unmount with the filter they remove,
+  // The feed's and Plex's active-filter chips unmount with the filter they remove,
  // and a keyboard press left the focus on <body>. The chip hands it on itself, for every caller.
  it("a removing chip pressed from the keyboard hands the focus to its neighbour", () => {
    function Row() {
@@ -70,7 +70,7 @@ describe("Chip", () => {
    expect(document.activeElement).toBe(screen.getByRole("button", { name: "Remove News" }));
  });

-  // Round 3: the neighbour it hands the focus to can go in the same removal — the "Clear all" that
+  // The neighbour it hands the focus to can go in the same removal — the "Clear all" that
  // only shows beside enough filters. The focus then goes to the chip before, not to <body>.
  it("a removing chip whose neighbour goes with it hands the focus to the chip before", async () => {
    function Row() {
@@ -58,7 +58,7 @@ export const Chip = forwardRef<HTMLButtonElement, ChipProps>(function Chip(
      className={clsx(chipClass(variant, size), removeLabel && "pr-1.5", className)}
      onClick={(e) => {
        // A removal chip unmounts with what it removes: from the keyboard it holds the focus, which
-        // went to <body> (round 2, F6 — the active-filter chips of the feed and Plex). Every
+        // went to <body> (the active-filter chips of the feed and Plex). Every
        // removal chip gets it here, once, instead of each caller remembering.
        if (removeLabel && e.currentTarget === document.activeElement)
          handOffFocus(e.currentTarget, () => onClick?.(e));
@@ -126,8 +126,8 @@ export function DismissibleChip({
        type="button"
        onClick={(e) => {
          // From the keyboard the × holds the focus, and the chip unmounts: hand the focus to the
-          // next chip (else the previous one, else what follows the row) first, not to <body> (E28
-          // review F5). A mouse press on it leaves the focus where it was.
+          // next chip (else the previous one, else what follows the row) first, not to <body>.
+          // A mouse press on it leaves the focus where it was.
          const chip = e.currentTarget.parentElement;
          if (chip && e.currentTarget === document.activeElement) handOffFocus(chip, onDismiss);
          else onDismiss();
@@ -1,7 +1,7 @@
 // @vitest-environment jsdom
 //
-// A clamped text offers More only when it is actually cut (E30 S6 on the person page, shared with the
-// collection page in the E30 review, where a one-line summary carried a More that did nothing).
+// A clamped text offers More only when it is actually cut (built for the person page, shared with
+// the collection page, where a one-line summary carried a More that did nothing).
 import { afterEach, beforeAll, describe, expect, it, vi } from "vitest";
 import { act, cleanup, fireEvent, render, screen } from "@testing-library/react";
 import { ClampedText } from "./ClampedText";
@@ -207,7 +207,7 @@ describe("Combobox", () => {
  });

  it("names a chip by a label that arrives with later options, in that same render", () => {
-    // E32 S3: the label cache was a ref an effect filled AFTER the render that needed it, and nothing
+    // The label cache was a ref an effect filled AFTER the render that needed it, and nothing
    // re-rendered afterwards — the chip kept showing the raw id until something unrelated did.
    const view = (options: { value: string; label?: string }[]) => (
      <FloatingTree>
@@ -298,12 +298,12 @@ describe("Combobox", () => {
  });
 });

-// PageUp/PageDown (E11, owner request): the sidebar's channel filter and the Plex pickers run to
+// PageUp/PageDown (owner request): the sidebar's channel filter and the Plex pickers run to
 // hundreds of rows, and floating-ui's list navigation has only the arrows. jsdom has no layout, so a
 // page is the fixed fallback of 10 rows; the measured page is pinned in lib/listPaging.
-// E28 review J1 + J2: a failed list said "No matches" (an error dressed as an empty result), and a
+// A failed list said "No matches" (an error dressed as an empty result), and a
 // picker whose pick was still being written took another pick.
-// E28 review F5: a chip's × pressed from the keyboard unmounted with the focus on it — to <body>,
+// A chip's × pressed from the keyboard unmounted with the focus on it — to <body>,
 // out of the picker. The focus goes to the next chip's × (else the previous one, else the field).
 describe("Combobox · removing a chip from the keyboard", () => {
  function Harness({ initial }: { initial: string[] }) {
@@ -335,7 +335,7 @@ describe("Combobox · honest states", () => {
    expect(screen.queryByText("common.noResults")).toBeNull();
  });

-  // A failed list with no way back but closing its dialog was a dead end (E28 review, owner go).
+  // A failed list with no way back but closing its dialog was a dead end.
  it("offers a Retry on the error, by click and by Enter in the field, keeping the focus", () => {
    const onRetry = vi.fn();
    wrap(
@@ -363,7 +363,7 @@ describe("Combobox · honest states", () => {
    expect(document.activeElement).toBe(field);
  });

-  // Round 2 (F4): a listbox may own options only — the alert, "Loading…", "No matches" and the
+  // A listbox may own options only — the alert, "Loading…", "No matches" and the
  // "n more" line sat inside it (axe: aria-required-children). They sit beside it now.
  it("keeps the lines about the list outside the listbox, which the field still controls", () => {
    const { rerender } = wrap(
@@ -15,7 +15,7 @@ import { ChipLabelPending } from "./Skeleton";

 // The universal combobox: a react-select / Overseerr-style "combi input" — selected values are
 // removable chips that live INSIDE a text field you type into, and a checklist of options opens below.
-// It is the ONE control for "pick value(s) from a (possibly large, possibly async) vocabulary": the E5
+// It is the ONE control for "pick value(s) from a (possibly large, possibly async) vocabulary": the
 // Plex filters (genre / age-rating / studio, multi-select) and the collection picker (single-select,
 // server-searched) are both instances, and the header search typeahead can move onto it later.
 //
@@ -56,12 +56,12 @@ export interface ComboboxProps {
  clientFilter?: boolean;
  loading?: boolean;
  /** The options failed to load: said in the list instead of "No matches" (an error is not an empty
-   *  result — E28 review J1). */
+   *  result). */
  error?: string;
  /** Loads the options again after `error`: a Retry beside the message, and Enter in the field. A
   *  failed list with no way back but closing the dialog was a dead end. */
  onRetry?: () => void;
-  /** Takes no pick (e.g. the last one is still being written, J2). Soft, like `FocusSafeButton`: the
+  /** Takes no pick (e.g. the last one is still being written). Soft, like `FocusSafeButton`: the
   *  field keeps the focus (a natively disabled one drops it to <body>), announced as unavailable. */
  disabled?: boolean;
  /** Shown when there are no options to list. */
@@ -81,7 +81,7 @@ export interface ComboboxProps {
 /** Remember each value's display label as options flow through, so a selected chip can name itself
 *  even after the option leaves the query-narrowed list (async single-select). State, merged DURING
 *  render: a ref written in an effect was read by the render before it, so a chip could name itself
- *  by its raw value until something else re-rendered (E32). */
+ *  by its raw value until something else re-rendered. */
 function useLabelCache(options: readonly ComboboxOption[]): Record<string, string> {
  const [labelCache, setLabelCache] = useState<Record<string, string>>({});
  const newLabels = options.filter(
@@ -126,7 +126,7 @@ export default function Combobox({
    onQueryChange?.(q);
  };
  // The input node as STATE, not a ref: the handlers below are handed to getReferenceProps during
-  // render, and a ref inside them made the React Compiler skip the whole component (E32).
+  // render, and a ref inside them made the React Compiler skip the whole component.
  const [inputEl, setInputEl] = useState<HTMLInputElement | null>(null);
  const { ref: fadeRef, style: fadeStyle } = useScrollFade();

@@ -294,7 +294,7 @@ export default function Combobox({
      >
        {/* The scroll box owns the height cap AND the scroll-fade mask (ref+style on the same element). */}
        <div ref={fadeRef} style={fadeStyle} className="max-h-64 overflow-y-auto no-scrollbar">
-          {/* The listbox holds the options alone; the lines about the list sit beside it (F4). */}
+          {/* The listbox holds the options alone; the lines about the list sit beside it. */}
          <PopoverList>
            {!showError &&
              visible.map((o) => {
@@ -124,7 +124,7 @@ describe("CommandPalette", () => {
    ).toContain("page:feed");
  });

-  // E28 F35: the active row wore `bg-card`, ~1.07:1 on the palette's glass. It is the menus' row now.
+  // The active row wore `bg-card`, ~1.07:1 on the palette's glass. It is the menus' row now.
  it("marks the active row with the menus' roving highlight, and only that row", () => {
    open();
    const rows = screen.getAllByRole("option");
@@ -174,7 +174,7 @@ describe("CommandPalette", () => {
  });
 });

-// The owner's case (E11 S3): typing "theme" put a channel matched only through its handle above every
+// The owner's case: typing "theme" put a channel matched only through its handle above every
 // "Theme: …" row, and Enter opened the channel.
 describe("CommandPalette · ranking", () => {
  const noop = () => {};
@@ -276,7 +276,7 @@ describe("CommandPalette · accessibility", () => {
  });
 });

-// Owner request (E11): with 461 rows and ~11 in view, the arrows alone were a slog. PageUp/PageDown
+// Owner request: with 461 rows and ~11 in view, the arrows alone were a slog. PageUp/PageDown
 // move by what fits in the list, and — unlike the arrows — stop at the ends instead of wrapping.
 describe("CommandPalette · paging", () => {
  const many = (): CommandGroup[] => [
@@ -9,7 +9,7 @@
 // and history.ts's overlay bookkeeping is module-global, so the sibling tests there leave an
 // in-flight programmatic back() behind — under which this one either hangs or, worse, goes green on
 // somebody else's popstate. That is exactly how it used to pass: it asserted nothing on its own, and
-// running it with `-t` failed (E10 review round 1). Its own file gives it clean module state.
+// running it with `-t` failed. Its own file gives it clean module state.
 import { useState } from "react";
 import { describe, it, expect, vi, beforeAll } from "vitest";
 import { render, screen, fireEvent, waitFor, act } from "@testing-library/react";
@@ -11,7 +11,7 @@ import { ROVING_HIGHLIGHT } from "./AnchoredPopover";
 import { fieldClass } from "./fieldStyles";

 // The app's jump-to bar: one text field over every place you can go and every app-wide thing you can
-// do, opened with Ctrl/Cmd+K. It is the "one-layer operation" half of E10 — reaching a channel three
+// do, opened with Ctrl/Cmd+K. It is the "one-layer operation" idea — reaching a channel three
 // modules away should cost a keystroke and a word, not a rail, a page and a scroll.
 //
 // Domain-free on purpose: the items arrive as a prop, because `components/ui` may not import a
@@ -58,7 +58,7 @@ const byRank = (a: Rank, b: Rank) => a[0] - b[0] || a[1] - b[1];
 *  the hint, the hidden keywords or the group name carry it. An item ranks by its worst word first, so
 *  a row whose label holds every word beats one that needed its handle for any of them. Before this
 *  the list was pure group order: "theme" put the channel "The Meltzone Podcast" — matched through its
- *  handle `@themeltzone` alone — above every "Theme: …" row, and Enter opened a channel (E11 S3). */
+ *  handle `@themeltzone` alone — above every "Theme: …" row, and Enter opened a channel. */
 function rankItem(item: CommandItem, groupLabel: string, words: string[]): Rank | null {
  const label = fold(item.label);
  const rest = fold(`${item.hint ?? ""} ${item.keywords ?? ""} ${groupLabel}`);
@@ -262,21 +262,21 @@ function CommandPaletteBody({ onClose, groups }: { onClose: () => void; groups:
                // The combobox input owns the tab stop; the arrows move a VIRTUAL selection over
                // these rows. Leaving them tabbable put DOM focus on a row while
                // aria-activedescendant named a different one — two competing "current" options —
-                // and made Tab walk every result (E10 review round 1).
+                // and made Tab walk every result.
                tabIndex={-1}
                onClick={() => choose(item)}
                onMouseMove={(e) => {
                  // Only a pointer that actually MOVED may take the selection. After the arrow keys
                  // scroll a row past the fold, Blink re-dispatches a mousemove at the unchanged
                  // cursor position to refresh hover — which handed the highlight straight back to
-                  // whatever row slid under a resting cursor (E10 review round 1). The browser
+                  // whatever row slid under a resting cursor. The browser
                  // reports zero movement for exactly that synthetic event, so ask it directly:
-                  // round 1 remembered the last position instead, which could not judge the FIRST
-                  // move of a session and went stale across a close (round 2).
+                  // remembering the last position instead could not judge the FIRST
+                  // move of a session and went stale across a close.
                  if (e.movementX === 0 && e.movementY === 0) return;
                  setPicked(item.id);
                }}
-                // The menus' roving highlight: a `bg-card` row was ~1.07:1 on the palette's glass (E28).
+                // The menus' roving highlight: a `bg-card` row was ~1.07:1 on the palette's glass.
                data-active={item.id === activeId ? "true" : undefined}
                className={`flex w-full items-center gap-2 rounded-lg px-2 py-1.5 text-sm text-muted hover:text-fg ${ROVING_HIGHLIGHT}`}
              >
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// Contract of the right-click / long-press context menu (E10 S2). Same floating-ui recipe as
+// Contract of the right-click / long-press context menu. Same floating-ui recipe as
 // ActionMenu.dom.test.tsx: stub the two observers autoUpdate wires and wrap in a FloatingTree, and
 // query with `{ hidden: true }` because floating-ui never positions in jsdom, so the surface stays
 // `visibility: hidden` and the a11y tree treats its contents as hidden.
@@ -176,7 +176,7 @@ describe("useContextMenu / ContextMenu", () => {

  // A card the keyboard is on is focused as a whole, through a tabindex=-1 root (lib/listNav), and
  // holds tabbable controls. floating-ui handed the focus back to the first of them instead of the
-  // card — measured in Chrome (deep-review E11 round 2): j, Shift+F10, Escape left it on the thumbnail.
+  // card — measured in Chrome: j, Shift+F10, Escape left it on the thumbnail.
  // Closed here by picking the row (the other way out; jsdom delivers no Escape to a keyboard-opened
  // menu — Escape is measured in the browser).
  it("gives the focus back to a card focused as a whole, not to a control inside it", async () => {
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// The R14 zoom-fit foundation: DataTable's opt-in `fixedLayout` renders a <colgroup> and puts the
+// The zoom-fit foundation: DataTable's opt-in `fixedLayout` renders a <colgroup> and puts the
 // table into `table-layout: fixed`, so columns with a width keep it and the rest (fluid text columns)
 // split the remainder + truncate instead of pushing the table wider than its container. Off by
 // default so existing content-sized tables are untouched.
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// DataTable's `keyboardNav` (E11 S4c): the rows are a keyboard list — `j`/`k` walk them and Enter
+// DataTable's `keyboardNav`: the rows are a keyboard list — `j`/`k` walk them and Enter
 // clicks the element a column marks `data-nav-open`. It used to render its rows twice (a table and,
 // below `md`, a card list, one hidden by CSS), so `j` past the last row could carry on into the hidden
 // copy. Now it mounts exactly one layout, chosen by the `md` media query.
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E31 S4d: the channel manager's "All" page (~400 rows) blocked one frame for 0.7–0.9 s on prod to
+// The channel manager's "All" page (~400 rows) blocked one frame for 0.7–0.9 s on prod to
 // mount, and every key in its search re-rendered every row. A long page now mounts its first rows at
 // once and the rest in small background steps (progressiveRows), and a step leaves the rows it holds
 // alone (a row is its own component, whose cells the React Compiler caches).
@@ -101,7 +101,7 @@ describe("DataTable — a long page mounts in steps", () => {
    await waitFor(() => expect(bodyRows()).toBe(150));
  });

-  // E31 S5: Channels turns the page to the row a priority pick moved and focuses it in the commit
+  // Channels turns the page to the row a priority pick moved and focuses it in the commit
  // that follows. That row was at index 40 of a 50-row page: not mounted yet, focus fell to <body>.
  it("mounts a revealed row in the render that turns to its page", () => {
    host = document.createElement("div");
@@ -125,7 +125,7 @@ describe("DataTable — a long page mounts in steps", () => {
    expect(host.querySelector("tbody")?.textContent).toContain("r 90");
  });

-  // E31 review 2: the reveal floor held for the page's lifetime. Channels never clears its `follow`, so
+  // The reveal floor once held for the page's lifetime. Channels never clears its `follow`, so
  // after one priority pick every later sort or search mounted every row up to that channel at once.
  it("lets a revealed row go once the reader moves to another view", () => {
    host = document.createElement("div");
@@ -270,7 +270,7 @@ export default function DataTable<T>({
  const [revealSeen, setRevealSeen] = useState(reveal?.seq);
  // The view the reveal was made for: the floor below holds only there. A caller keeps its reveal
  // after the focus (Channels never clears `follow`), and a floor that outlived its view mounted every
-  // row up to that channel on each later sort or search (E31 review 2).
+  // row up to that channel on each later sort or search.
  const [revealView, setRevealView] = useState<string | null>(null);
  if (reveal && reveal.seq !== revealSeen) {
    setRevealSeen(reveal.seq);
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3d4 (B4): below ~600px of height a band's filter row folds into one "Filters · n" button that
+// DESIGN.md B4: below ~600px of height a band's filter row folds into one "Filters · n" button that
 // shows it again in place; on a tall window there is no button and the row is always there.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -38,7 +38,7 @@ describe("a selectable button", () => {
    expect(onClick).toHaveBeenCalledTimes(1);
  });

-  // E28 review H2: Enter / Space on a focused title with its words still selected (selected, then
+  // Enter / Space on a focused title with its words still selected (selected, then
  // tabbed to) did nothing — a keyboard click ends no selection; it carries no pointer at all.
  it("a keyboard click opens even with the title's words selected", () => {
    const onClick = vi.fn();
@@ -12,7 +12,7 @@ import { endsTextSelection, movedPastSlop } from "../../lib/textSelection";
 // A `<button>` whose `disabled` never throws the focus out of the page.
 //
 // A focused button that turns natively disabled loses the focus — Chrome moves it to <body>
-// (measured, E11 S2) — and the app disables the very button that was just pressed on ~60 sites (a
+// (measured) — and the app disables the very button that was just pressed on ~60 sites (a
 // mutation pending, a form gone clean after its save), so a keyboard user was dropped out of the page
 // on every submit. While THIS button holds the focus, `disabled` (and `busy`) is expressed as
 // `aria-disabled` plus a click guard instead: still focusable, announced as unavailable, and not
@@ -69,7 +69,7 @@ export const FocusSafeButton = forwardRef<
  // not focused (another app in front) `.focus()` moves `activeElement` but fires no focus event
  // (measured in Chrome), and the button disabled in that state lost the focus all the same. Read
  // through useSyncExternalStore, which React asks on EVERY render, never from a memo: a bare
-  // render-time DOM read made the React Compiler skip the component (E32), and compiled it would be
+  // render-time DOM read made the React Compiler skip the component, and compiled it would be
  // cached. Nothing to subscribe to — the focus events above already re-render it.
  const activeNow = useSyncExternalStore(
    noSubscription,
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 review F2: a column filter is shown only by its column's header funnel, so a filter set on a
+// A column filter is shown only by its column's header funnel, so a filter set on a
 // column that is off screen — a `tabletHidden` column on a 1024 window, any column in the card layout
 // — narrowed the rows with nothing on screen saying so. The cue names it, and its chip removes it.
 import { afterEach, describe, expect, it, vi } from "vitest";
@@ -9,7 +9,7 @@ import {
 } from "../../lib/tableColumns";
 import { Chip } from "./Chip";

-// E28 F2: a column filter lives in its column's header funnel, so when the column is off screen — a
+// A column filter lives in its column's header funnel, so when the column is off screen — a
 // `tabletHidden` / `narrowHidden` column below its width, or any column in a card layout, which has no
 // header at all — the rows are narrowed by something nobody can see ("12 of 392", and no why). This
 // names each such filter above the rows, and its chip removes it (owner's pick A, 2026-09-27).
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3b: the glyph-only button (DESIGN.md A2) and the two other click primitives that sit beside
+// The glyph-only button (DESIGN.md A2) and the two other click primitives that sit beside
 // `Button`: `Pressable` (a surface whose content is the control) and `DragHandle`.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3c: the field primitives (docs/DESIGN.md A4-A6): one frame per rung, the native controls handing
+// The field primitives (docs/DESIGN.md A4-A6): one frame per rung, the native controls handing
 // back values, and the dropdown trigger by context.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -2,7 +2,7 @@ import clsx from "clsx";
 import type { InputHTMLAttributes, Ref, TextareaHTMLAttributes } from "react";
 import { fieldClass, type FieldSize } from "./fieldStyles";

-// The text-field primitives (docs/DESIGN.md A4, E28 S3c). Every text, number, password, e-mail, search
+// The text-field primitives (docs/DESIGN.md A4). Every text, number, password, e-mail, search
 // and date box is an `Input`; a multi-line one a `Textarea`; a box whose frame also holds an icon,
 // chips or a clear button is a wrapper painted with `fieldClass` around an `InnerInput`. A raw
 // `<input>` or `<textarea>` anywhere else is a lint error (`local/input-primitive`), so the
@@ -71,7 +71,7 @@ describe("ItemCount", () => {
    expect(screen.getByText("5 items").className).toContain("opacity-60");
  });

-  // E31 review 2: the dim and aria-busy were the only stale cue, and a screen reader voices neither on
+  // The dim and aria-busy were once the only stale cue, and a screen reader voices neither on
  // plain text; a failed count looked like a first load.
  it("tells a screen reader a stale count is updating", () => {
    render(<ItemCount count={5} stale />);
@@ -8,7 +8,7 @@ import { formatNumber } from "../../lib/format";
 const OVERFLOW_AT = 1_000_000;

 // A compact, FIXED-WIDTH count pill for a toolbar: the filtered item count at the toolbar's 28, flat
-// and tonal like a `Badge` — no edge, because it does nothing (E28 S3d: edged = pressable, flat = just
+// and tonal like a `Badge` — no edge, because it does nothing (edged = pressable, flat = just
 // says; it used to wear the filter dropdowns' glass pill and read as one more). The fixed width is the point — a
 // changing number must never shove the controls beside it, so the box is sized to the widest value
 // ("999,999 items") and the content is centred within it. A count of a million or more collapses to
@@ -17,7 +17,7 @@ const OVERFLOW_AT = 1_000_000;
 // kept while the new one loads (keepPreviousData) — dimmed, and "updating" for a screen reader (which
 // does not voice aria-busy on plain text), so a broad search's slow count ("a": ~3 s) never reads as
 // the new result. `failed`: the count could not be read — a dash that says so, not the icon-only box
-// of a first load (E31 review 2).
+// of a first load.
 export default function ItemCount({
  count,
  stale = false,
@@ -98,9 +98,9 @@ describe("KeyboardHelp", () => {

  // The context menu is NOT app-wide: it has its own scope, registered by `useContextMenu`, so the
  // sheet only offers Shift+F10 where something actually answers it. Advertising it globally was a
-  // promise the feed, Plex, Downloads and Settings did not keep (E10 review round 1) — the test above
+  // promise the feed, Plex, Downloads and Settings did not keep — the test above
  // pins the other half, that a bare app lists the two global keys and nothing else.
-  // Through the REAL hook, not a scope stub: round 2 showed that commenting out
+  // Through the REAL hook, not a scope stub: commenting out
  // `useKeyScope("contextMenu")` inside `useContextMenu` left the whole suite green, because the
  // only coverage registered the scope by hand. This binds the sheet to the thing that owns it.
  it("offers the context-menu key on a surface that has one, and not otherwise", () => {
@@ -42,7 +42,7 @@ export default function KeyboardHelp() {
  // them there would promise keys that do nothing until the overlay closes. And a `g` chord is listed
  // only for a module this account can actually reach.
  // The context-menu key goes too: every `useContextMenu` is a page list, and with an overlay up the
-  // focus Shift+F10 needs is inside the overlay (pre-existing since E10, found in S4 UAT).
+  // focus Shift+F10 needs is inside the overlay (found in UAT).
  const live = overLayer
    ? scopes.filter((s) => !isPageLevelScope(s) && s !== "contextMenu")
    : scopes;
@@ -7,7 +7,7 @@ export type MatchMode = "any" | "all";
 // a captioned segment on the panel, right ABOVE its picker — never in the picker's popover, where a
 // closed picker hid that "all" was on (the owner, 2026-09-27; DESIGN "Action placement": the state
 // that shapes the results stays in sight). With fewer than two picks any and all are the same filter,
-// so it draws nothing then. The panel's captioned toggle is the one 24px segment (DESIGN log, S3g).
+// so it draws nothing then. The panel's captioned toggle is the one 24px segment (DESIGN log).
 export default function MatchToggle({
  picked,
  value,
@@ -5,7 +5,7 @@ import { iconButtonClass } from "./buttonStyles";
 import { MoreButton } from "./MoreButton";
 import { SURFACE } from "./surfaces";

-// The ONE media card (docs/DESIGN.md "Media card", E28 S3h): a YouTube video, a Plex movie, show,
+// The ONE media card (docs/DESIGN.md "Media card"): a YouTube video, a Plex movie, show,
 // season or episode — in the feed, the Plex library, Home's shelves and the media pages' strips — is
 // the same object: one glass frame holding the picture, then the title, then a meta line PINNED to the
 // frame's bottom edge. Cards in a grid row stretch to the row's height, so pinning the meta puts every
@@ -3,7 +3,7 @@ import { PopoverSurface } from "./AnchoredPopover";
 import type { useMediaPreview } from "./useMediaPreview";
 import { SURFACE } from "./surfaces";

-// A media card's preview (docs/DESIGN.md "Media card", E28 S3h — the owner's pick C, 2026-09-26):
+// A media card's preview (docs/DESIGN.md "Media card" — the owner's pick C, 2026-09-26):
 // rest the mouse on a card for a moment and a panel beside it shows what the card has no room for —
 // the whole title, the facts, the synopsis in lines short enough to read. It replaced the synopsis as
 // the picture's native tooltip, which fired anywhere on the art, ran one long line, and shared its
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// Modal: focus management (FE-9). This was DEFERRED from the E4 deep-review because it's
+// Modal: focus management (FE-9). This was DEFERRED from a code review because it's
 // jsdom-untestable as written — `focusablesIn` filters on `el.offsetParent !== null`, and jsdom
 // returns null for EVERY element (it never lays out), so the trap would see zero focusables and do
 // nothing. We stub `HTMLElement.prototype.offsetParent` to report "visible" (non-null) for any
@@ -225,7 +225,7 @@ describe("Modal focus trap (offsetParent stubbed 'visible')", () => {
    // The Tab trap deliberately spans the dialog ROOT so a toast's Undo/✕ is reachable; ownership
    // was tracked over the PANEL, so touching that ✕ read as "the user left the dialog" and the
    // restore was skipped — and removing the toast fires no focusin to re-arm it. Measured landing
-    // on <body> (deep-review round 6).
+    // on <body>.
    const trigger = document.createElement("button");
    document.body.appendChild(trigger);
    trigger.focus();
@@ -281,7 +281,7 @@ describe("Modal focus trap (offsetParent stubbed 'visible')", () => {
    // layer is above me" (measured). So a dialog that closes on its own — an in-flight mutation's
    // `onSuccess`, not a click, since a scrim covers it — under a modal raised above it yanked the
    // focus out of that standing dialog onto its own trigger, behind the scrim, where Enter still
-    // activated it (deep-review round 7). Sibling dialogs, not nested: that is the arrangement
+    // activated it. Sibling dialogs, not nested: that is the arrangement
    // `ConfirmProvider` and the app-wide `ErrorDialog` actually use.
    const trigger = document.createElement("button");
    document.body.appendChild(trigger);
@@ -366,7 +366,7 @@ describe("Modal — toasts raised while it is open", () => {
  });
 });

-// E28: one Escape in a dialog's open list closed the list AND the dialog (the share dialog's user
+// One Escape in a dialog's open list closed the list AND the dialog (the share dialog's user
 // picker). The list's `document` listener closes it with a synchronous flush, which pops its layer
 // before the dialog's `window` listener runs — so "am I topmost?" reads yes. The dialog must honour
 // the list's stamp, as the players already do (UX-8).
@@ -4,7 +4,7 @@
 // <body> is painted beneath it, i.e. invisible). Leaving fullscreen changes that container — and a
 // React portal whose container changes is torn down and recreated, so everything the dialog held went
 // with it: a half-filled download form reset to its defaults and focus fell to <body> (a verifier's
-// side finding in deep-review round 4).
+// side finding in a review).
 import { describe, it, expect, afterEach } from "vitest";
 import { render, cleanup, screen, fireEvent, act } from "@testing-library/react";
 import { vi } from "vitest";
@@ -58,7 +58,7 @@ describe("a dialog with an autoFocus child", () => {
    // has already called `.focus()` on a disconnected node — a no-op — and Modal's own fallback
    // then puts the focus on the first focusable in the panel, the header's ✕. That is what the
    // command palette, the download rename dialog and every non-danger confirm shipped with after
-    // deep-review round 4 (found in round 5).
+    // the first fix (found by a later review).
    render(
      <Modal initialFocus="close" title="Search" onClose={() => {}}>
        <input aria-label="query" autoFocus />
@@ -4,11 +4,11 @@
 // mount→cleanup→mount is what the code under test reacts to, and the module-level layer stack in
 // ui/Modal (and the overlay-host registry in ui/Toaster) leaks between tests in one file.
 //
-// Round 5 fixed this in Chrome and shipped it with NO test: the second mount found the focus
+// A fix settled this in Chrome and shipped it with NO test: the second mount found the focus
 // outside the dialog (its own cleanup had just handed it back to the trigger) and moved it to the
 // first focusable — taking it off an `autoFocus` field a beat after the browser had put it there.
 // The command palette opened focused on its ✕ in dev and on its search field in prod, from the same
-// code. Round 6 measured that deleting the fix left all 1009 tests green, which is why this exists.
+// code. A check measured that deleting the fix left all 1009 tests green, which is why this exists.
 //
 // Same offsetParent stub as Modal.dom.test.tsx: jsdom never lays out, so every element reports
 // offsetParent null and `focusablesIn` would see nothing.
@@ -52,8 +52,8 @@ const PANEL = "relative flex flex-col overflow-hidden outline-none panel-scroll-

 // Small centered modal shell: backdrop + ESC + scroll-lock close. Each open modal is a layer in the
 // shared `layerStack` so ESC only closes the topmost one — e.g. an error dialog over the tag editor
-// dismisses just the error and returns to the editor underneath, and a player below defers to it
-// (R8 S3). Portals into `document.fullscreenElement ?? document.body`: a node under <body> is painted
+// dismisses just the error and returns to the editor underneath, and a player below defers to it.
+// Portals into `document.fullscreenElement ?? document.body`: a node under <body> is painted
 // beneath whatever holds the fullscreen lock, so a Modal opened from a player in true fullscreen (the
 // download dialog) would otherwise open invisibly.
 export default function Modal({
@@ -85,15 +85,15 @@ export default function Modal({
  // Browser/mouse Back closes the topmost modal instead of navigating away.
  useBackToClose(onClose);
  // ONE container element for the dialog's whole life, re-parented between <body> and the fullscreen
-  // element (deep-review round 4) — and attached before the children commit, so an `autoFocus` child
-  // still gets the focus (round 5). Both rules live in the hook; `ui/AnchoredPopover` shares it.
+  // element — and attached before the children commit, so an `autoFocus` child
+  // still gets the focus. Both rules live in the hook; `ui/AnchoredPopover` shares it.
  const host = useFullscreenPortalHost();
  const panelRef = useRef<HTMLDivElement>(null);
  const rootRef = useRef<HTMLDivElement>(null);
  const titleId = useId();
  // Focus lands in the panel; OWNERSHIP is the whole dialog — the same element the Tab trap below
  // is scoped to, toast host included. Tracking ownership over the panel alone meant tabbing to an
-  // in-dialog toast counted as leaving, and the close-time restore was then skipped (round 6).
+  // in-dialog toast counted as leaving, and the close-time restore was then skipped.
  const bodyRef = useRef<HTMLDivElement>(null);
  const closeRef = useRef<HTMLButtonElement>(null);
  const [footerSlot, setFooterSlot] = useState<HTMLDivElement | null>(null);
@@ -108,11 +108,11 @@ export default function Modal({
  // to take the focus at open, so it lands on the ✕. When the form arrives the focus moves to its
  // first field — but only while it still rests on the ✕: a user who has moved it keeps it there.
  // The form takes it ONCE: arrived while the focus was elsewhere, it disarms all the same, or a later
-  // mutation would pull the focus off a ✕ the user went back to (E32 review round 2). The watch
+  // mutation would pull the focus off a ✕ the user went back to. The watch
  // re-arms only when the focus is taken away, not moved: the body has no control again, or the
  // focused one went (a failed load's Retry giving way to the spinner, or straight to the form) and
  // the focus fell to <body>. It then goes to the ✕, and on to the first field as soon as the body
-  // has one — at once, when other controls stayed beside the one that went (E32 review). Only
+  // has one — at once, when other controls stayed beside the one that went. Only
  // <body>: focus in a dialog stacked on top is not this one's to take.
  useEffect(() => {
    const body = bodyRef.current;
@@ -1,7 +1,7 @@
 import clsx from "clsx";
 import type { InputHTMLAttributes, ReactNode, Ref } from "react";

-// A6 (docs/DESIGN.md, E28 S3c): checkbox, slider and colour well are the BROWSER's controls, tinted
+// A6 (docs/DESIGN.md): checkbox, slider and colour well are the BROWSER's controls, tinted
 // with the accent — keyboard, screen reader and forced-colours support come free, and a drawn copy
 // would have to rebuild all three. What the browser does not give is the 44 touch target: the labelled
 // row (`CheckRow`) or the slider's own box grows to it on a coarse pointer. Like `Input`, a raw
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3d (B7): the nav rail's row. Icon-only, its name is the module's full name (with an unread
+// DESIGN.md B7: the nav rail's row. Icon-only, its name is the module's full name (with an unread
 // count, which was only ever drawn); with the labels on, the name is the short label the user SEES
 // (WCAG 2.5.3), the full one staying in the tooltip. "You are here" is aria-current plus the tint.
 import { readFileSync } from "node:fs";
@@ -49,7 +49,7 @@ describe("NavRailItem", () => {
    expect(row.getAttribute("aria-current")).toBeNull();
  });

-  // E28 audit: labelled, the name was the row's text, the corner badge first, so 20 unread was heard
+  // Audit: labelled, the name was the row's text, the corner badge first, so 20 unread was heard
  // as "9+Alerts" where the icon-only row says "Notifications (20)".
  it("labelled with a count: the visible label, then the exact count", () => {
    render(
@@ -66,7 +66,7 @@ describe("NavRailItem", () => {
    expect(screen.getByRole("link", { name: "Alerts (20)" })).toBeTruthy();
  });

-  // E31 (the owner): a middle-click on a rail row opens the page in a new tab. The row is a real link
+  // Owner's call: a middle-click on a rail row opens the page in a new tab. The row is a real link
  // to the page, so the browser's own gestures work; only a plain left click navigates in-app.
  it("is a link to its page, navigating in-app only on a plain left click", () => {
    const onClick = vi.fn();
@@ -124,7 +124,7 @@ describe("NavRailAction", () => {
  });
 });

-// E31 (the owner's pick): the five admin rows sat below the fold at 133% zoom behind a fade alone, so
+// The owner's pick: the five admin rows sat below the fold at 133% zoom behind a fade alone, so
 // they are one "Admin" row opening a panel of them. The panel's pages stay links, as the rail's are.
 describe("NavRailGroup", () => {
  const pages = (active: string | null, onSelect = vi.fn()) =>
@@ -92,7 +92,7 @@ export function NavRailItem({
  active?: boolean;
  /** An unread / in-progress count on the icon's corner (0 hides it). */
  badge?: number;
-  /** The page's own address. A page row is a real link (the owner, E31): a middle-click, a
+  /** The page's own address. A page row is a real link (the owner's call): a middle-click, a
   *  Ctrl/Cmd-click or the browser's "Open in new tab" opens the page in a tab of its own. */
  href: string;
  /** The in-app navigation, run only for a plain left click. */
@@ -186,7 +186,7 @@ export interface NavRailGroupPage {
  onSelect: () => void;
 }

-/** One rail row standing for a set of rare pages (E31, the owner's pick: the admin pages), opening a
+/** One rail row standing for a set of rare pages (the owner's pick: the admin pages), opening a
 *  panel of them to its right. The rail fits a 133–150% zoom without a scroll nobody sees: five admin
 *  rows sat below the fold behind a fade alone. The row is "you are here" while any of its pages is.
 *  The panel's pages are real links, as the rail's own rows are (a middle-click opens a tab), so it is a
@@ -4,7 +4,7 @@ import { createPortal } from "react-dom";
 // The app's overlay-root convention. Any full-screen or floating overlay renders THROUGH this so it
 // escapes whatever ancestor stacking context it was declared in — a `mask-image`/`transform`/`filter`
 // on the page scroller (or a rail's `backdrop-filter`) would otherwise trap a `fixed` child and let
-// the chrome paint over it. That is the exact class of bug that buried the players in E3; the
+// the chrome paint over it. That is the exact class of bug that once buried the players; the
 // modal/player family already dodges it by portaling to <body>, and this makes that the one blessed
 // way. Stacking among overlays is then decided by DOM order + the z-index token scale (tailwind
 // `zIndex`), never by an ad-hoc magic number. Portals to <body> (the existing de-facto target).
@@ -81,7 +81,7 @@ describe("PageScroller scroll memory", () => {
    expect(main.scrollTop).toBe(200);
  });

-  // Round 2 (F8): the page's very end, whose Back-to-top spacer only shows once scrolled, is out of
+  // The page's very end, whose Back-to-top spacer only shows once scrolled, is out of
  // reach when Back rebuilds the page — and the restore gave up with the page at the top.
  it("Back to a spot within the end's spacer lands at the page's end, not its top", () => {
    setEntry(1);
@@ -125,7 +125,7 @@ describe("PageScroller scroll memory", () => {
    expect(main.scrollTop).toBe(450);
  });

-  // Round 3: landing at the end shows Back to top, whose spacer then brings the spot within reach.
+  // Landing at the end shows Back to top, whose spacer then brings the spot within reach.
  it("finishes the way to a spot the spacer brings within reach after it lands", () => {
    let frames: FrameRequestCallback[] = [];
    vi.stubGlobal("requestAnimationFrame", (cb: FrameRequestCallback) => frames.push(cb));
@@ -154,7 +154,7 @@ describe("PageScroller scroll memory", () => {
    expect(main.scrollTop).toBe(500);
  });

-  // Round 3: the give-up put the page at its end 1.5 s in, wherever the reader had scrolled to.
+  // The give-up put the page at its end 1.5 s in, wherever the reader had scrolled to.
  // Every way a reader takes the page: a wheel, a finger, the scrollbar (a pointer), a key.
  it.each([
    ["a wheel", (main: HTMLElement) => fireEvent.wheel(main)],
@@ -213,7 +213,7 @@ describe("PageScroller scroll memory", () => {
  });
 });

-// E28 UAT: Back to top floats over the bottom-right corner, and at the end of a long table it covered
+// UAT: Back to top floats over the bottom-right corner, and at the end of a long table it covered
 // the closing pager's "100" and "All". While it is up, the page ends a button's height later.
 describe("PageScroller · room for Back to top", () => {
  it("ends the page later while the button is up, and not otherwise", () => {
@@ -37,10 +37,10 @@ function rememberScroll(key: string, top: number): void {
 // clamp short. Re-apply over a few frames until it sticks (or the content is simply shorter now).
 // A spot the page can no longer reach — the last stretch of a page scrolled to its very end, whose
 // Back-to-top spacer only shows once scrolled — goes as far as the page does: it used to be left at
-// the top (round 2, F8). Only a spacer's worth short, and only once the height has held still, is
+// the top. Only a spacer's worth short, and only once the height has held still, is
 // that answer early; a list still loading is further short than that, and gets the full wait. And
 // landing there is not the end of it: that scroll is what shows Back to top, whose spacer then
-// makes the spot reachable, so the loop watches a few frames more and finishes the way (round 3).
+// makes the spot reachable, so the loop watches a few frames more and finishes the way.
 //
 // The reader scrolling meanwhile — a wheel, a finger, a key, the scrollbar — takes the page over:
 // the restore stops where they are, instead of pulling them back or dropping them at the end.
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E5 S7: PageShell measures the floating header and publishes its real height as `--hdr-h` onto the
+// PageShell measures the floating header and publishes its real height as `--hdr-h` onto the
 // content wrapper, so a header that flex-wraps to 2–3 rows on a narrow window never overlaps the
 // content below it. jsdom does no layout (offsetHeight is 0) and has no ResizeObserver, so we stub
 // both: a ResizeObserver that hands us its callback, and offset metrics on the measured header.
@@ -32,7 +32,7 @@ afterEach(() => {
  vi.unstubAllGlobals();
 });

-describe("PageShell --hdr-h (E5 S7)", () => {
+describe("PageShell --hdr-h", () => {
  it("publishes the measured header height onto the content wrapper", () => {
    const { container } = render(
      <PageShell header={<div data-testid="hdr">H</div>}>content</PageShell>,
@@ -139,7 +139,7 @@ describe("PageShell --rail-pinned-w / --content-mw", () => {
  });
 });

-// E28 S3d4 (B4): every page has one h1 — the page's name, visually hidden, unless a view shows its
+// DESIGN.md B4: every page has one h1 — the page's name, visually hidden, unless a view shows its
 // own title (a channel, a Plex show), which is then the page's h1 instead.
 describe("PageShell page heading", () => {
  it("names the page with a visually hidden h1", () => {
@@ -111,7 +111,7 @@ export default function PageShell({
  }, [maxWidth]);
  // The floating header pads the content below it by `--hdr-h`. That must equal the header's REAL
  // height, which grows as the flex-wrap header wraps to 2–3 rows on a narrow window — a fixed pad
-  // would let the content toolbar slide up under it (E5 S7). Measure the header and write its height
+  // would let the content toolbar slide up under it. Measure the header and write its height
  // straight onto the content wrapper that consumes `--hdr-h`.
  //
  // Both nodes are held in STATE (callback refs), NOT plain refs behind a one-shot `useLayoutEffect([])`:
@@ -34,7 +34,7 @@ describe("Pager · rows per page", () => {
  });
 });

-// The page box DERIVES the current page (E27 S2): it holds a draft only while being edited, so the
+// The page box DERIVES the current page: it holds a draft only while being edited, so the
 // parent moving the page needs no re-sync effect. The layout-phase probe is what proves "derived"
 // rather than "re-synced after paint" — a passive effect could not have run yet when it reads.
 describe("Pager · the page box follows the page without an effect", () => {
@@ -62,7 +62,7 @@ describe("PanelGroups — reordering without a drag", () => {
  });
 });

-// E28 review F4: every group's grip was "Drag to reorder" — a screen reader heard the same name five
+// Every group's grip was "Drag to reorder" — a screen reader heard the same name five
 // times and could not tell which group it would lift. It names its group, as the playlist grips do.
 describe("PanelGroups — the drag grip", () => {
  it("is named by its group", () => {
@@ -73,7 +73,7 @@ export function LoadingState({
 }

 // Centered inline status line for a page/section's loading and error(+retry) states, in the house
-// style (mirrors Playlists' error block). Text-only by design — R3's goal is to stop pages
+// style (mirrors Playlists' error block). Text-only by design — the goal is to stop pages
 // rendering an error or a still-loading fetch AS an "empty" state, not to add skeletons. The
 // `role=status`/`aria-live` also gives screen readers the state change (the app had zero live
 // regions before).
@@ -118,6 +118,6 @@ export function StateMessage({

 // The honest-state convention every page now follows: check `isError && <no data>` BEFORE the empty
 // branch, so a failed fetch renders this message (with Retry) instead of the page's "empty" text
-// — the R3 bug where a backend restart read as "No channels". The `&& <no data>` guard matters: a
+// — the bug where a backend restart read as "No channels". The `&& <no data>` guard matters: a
 // transient refetch error while data is already on screen keeps the data rather than blanking it to
 // an error. `StateMessage` is the shared building block pages wire into their loading/empty ternary.
@@ -32,7 +32,7 @@ describe("SegmentedToggle", () => {
    expect(onChange).toHaveBeenCalledWith("and");
  });

-  it("sits on the toolbar rung unless a panel's caption asks for the small one (S3g)", () => {
+  it("sits on the toolbar rung unless a panel's caption asks for the small one", () => {
    render(<SegmentedToggle options={OPTS} value="or" onChange={vi.fn()} ariaLabel="Default" />);
    const plain = screen.getByRole("group", { name: "Default" });
    expect(plain.className).toContain("text-xs");
@@ -39,7 +39,7 @@ export default function SegmentedToggle<T extends string>({
  fullWidth?: boolean;
  // md (28, text-xs) is the toolbar rung: it lines up with the search pill, the triggers and an `sm`
  // Button (DESIGN B1). sm (24, text-3xs) is only a side panel's captioned match toggle. md is the
-  // default because the old sm default slid a pager and a toolbar count onto 10px type (S3g).
+  // default because the old sm default slid a pager and a toolbar count onto 10px type.
  size?: "sm" | "md";
  className?: string;
 }) {
@@ -4,7 +4,7 @@ import clsx from "clsx";
 import { ChevronDown } from "lucide-react";
 import { FocusSafeButton } from "./FocusSafeButton";

-// The dropdown trigger (docs/DESIGN.md A5, E28 S3c): the one button that opens a list of choices —
+// The dropdown trigger (docs/DESIGN.md A5): the one button that opens a list of choices —
 // `SortControl`, `FilterSelect`, the view switcher, the date filter. By context: a pill at the toolbar's
 // 28 (in a toolbar the pill means "a filter or a choice"), or r8 at the form's 32, full width, lining up
 // with the inputs around it. The form trigger is one of the field family, so it wears the field's frame
@@ -45,8 +45,8 @@ export const SelectTrigger = forwardRef<
      {icon}
      {label != null && <span className="shrink-0 text-muted">{label}:</span>}
      {/* The toolbar value is a row, whatever the caller puts in it: an <svg> is display:block under the
-          preflight, and in a plain span it broke the value onto its own line (the saved-view chip,
-          E28). The form value is text that truncates, which a flex box would not do — nor would a bare
+          preflight, and in a plain span it broke the value onto its own line (the saved-view chip).
+          The form value is text that truncates, which a flex box would not do — nor would a bare
          text node in the toolbar row, so a text value gets its own truncating span: in a narrow place
          (the playlists rail) a long value ("Rendezés: Egyéni sorrend") used to wrap out of the pill. */}
      <span
@@ -84,7 +84,7 @@ describe("SidePanel · a mouse on a wide window", () => {
  });
 });

-// E28 review F6: the pinned (or peeking) card covers the tab, which only goes transparent — yet it
+// The pinned (or peeking) card covers the tab, which only goes transparent — yet it
 // stayed a Tab stop, an invisible control the focus landed on, named just "Unpin panel".
 describe("SidePanel · the tab under an open card", () => {
  it("is out of the Tab order while the card is open, and names its panel", () => {
@@ -79,7 +79,7 @@ export default function SidePanel({
  // adjusted during render, so it cannot come back open on the next turn.
  if (drawerOpen && (!drawerMode || pinned)) setDrawerOpen(false);
  // A key landing on the tab peeks the card open over it, and the tab then goes inert below — with
-  // the focus still on it: a hidden control, out of the tree, that Enter still pinned (round 2, F1).
+  // the focus still on it: a hidden control, out of the tree, that Enter still pinned.
  // The focus steps into the card as it opens, onto its pin: the same control. In the commit that
  // mounts the card, not a frame later — a frame never comes in a page the browser is not painting.
  useLayoutEffect(() => {
@@ -156,7 +156,7 @@ export default function SidePanel({
          nowhere. It only goes transparent: the card's rounded corner cuts away its own pixels, and
          this square-cornered tab sits right under that arc, so it would otherwise bleed through.
          `opacity-0` (not `invisible`/`hidden`) keeps it hoverable, and a focus already on it stays.
-          But no Tab lands on it while the card covers it (E28 review F6): an invisible stop, where the
+          But no Tab lands on it while the card covers it: an invisible stop, where the
          card's own pin says the same. A keyboard pin hands the focus into the card it opened. */}
      <Pressable
        {...(drawerMode ? {} : peek.hoverHandlers)}
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 S3d4 (UX-4): "Skip to content" moves the focus to the page's <main> itself. Following its
+// "Skip to content" moves the focus to the page's <main> itself. Following its
 // `#content` fragment would push a history entry, and the app reads every popstate as a page change.
 import { afterEach, describe, expect, it } from "vitest";
 import { cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E31 (the owner's pick for the Statistics overview): a figure's name sits by its number in a tile, and a
+// The owner's pick for the Statistics overview: a figure's name sits by its number in a tile, and a
 // breakdown is horizontal bars, largest first, so what spends the most is seen before a number is read.
 import { afterEach, describe, expect, it } from "vitest";
 import { cleanup, render, screen } from "@testing-library/react";
@@ -2,7 +2,7 @@ import type { ReactNode } from "react";
 import { HintLabel } from "./form";
 import { SURFACE } from "./surfaces";

-// E31 (the owner's pick for the Statistics overview): label…value rows stretched across a wide column
+// The owner's pick for the Statistics overview: label…value rows stretched across a wide column
 // made the eye travel from a name to its number. A tile keeps the two together, a row of them is read at
 // a glance, and a breakdown is a set of horizontal bars, so what spends the most is seen before a number
 // is read.
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 review F3: the tab rail was a row of `aria-pressed` toggle buttons — every tab a Tab stop, no
+// The tab rail was a row of `aria-pressed` toggle buttons — every tab a Tab stop, no
 // arrow keys, and nothing tying a tab to the content it shows. It is an ARIA tab list now.
 import { useState } from "react";
 import { afterEach, describe, expect, it } from "vitest";
@@ -9,12 +9,12 @@ import CountBadge from "./CountBadge";
 // Below it a side rail leaves a phone too little width for the content beside it (measured at 390 px:
 // the rail clipped its own labels and Settings → Account pushed its buttons over their text), so there
 // it is a horizontal row that scrolls. A page whose sub-tab is worth keeping persists it with
-// lib/storage's useTabPersistedState (per browser tab, E31 S4b: Admin Users, Config, Settings, Stats):
+// lib/storage's useTabPersistedState (per browser tab: Admin Users, Config, Settings, Stats):
 // a reload keeps the user where they were, and a fresh tab opens on the first sub-tab. Not every
 // caller: the channel manager's view is account-wide (ChannelsProvider), and Downloads, a channel page
 // and the player settings start on their first tab each time.
 //
-// An ARIA tab list (E28 review F3), not a row of toggle buttons: `tablist` / `tab` / `aria-selected`,
+// An ARIA tab list, not a row of toggle buttons: `tablist` / `tab` / `aria-selected`,
 // ONE Tab stop (the selected tab; roving tabindex), the arrows move and select (either axis, since the
 // vertical rail is a row below `sm`), Home/End jump to the ends. The panel it controls is wired with
 // `useTabPanel`: the tabs point at it (`aria-controls`) and it is labelled by the selected tab.
@@ -27,7 +27,7 @@ export interface TabDef {
  badge?: number; // optional count pill (e.g. pending requests); hidden when 0/undefined
  icon?: LucideIcon; // optional leading icon (the vertical settings rail)
  testId?: string; // optional data-testid, to preserve existing e2e hooks
-  /** The vertical rail's heading this tab sits under (E31: Configuration's sixteen groups under four).
+  /** The vertical rail's heading this tab sits under (Configuration's sixteen groups under four).
   *  Drawn above the first tab of each run; tabs of one heading stay together. */
  group?: string;
 }
@@ -95,7 +95,7 @@ describe("Toaster — handing the toasts to another host", () => {
  });

  it("takes the inherited set ONCE, at the takeover: a toast of its own stays an alert", () => {
-    // E32: the handover is a snapshot made when the host becomes top, not a read of what the top
+    // The handover is a snapshot made when the host becomes top, not a read of what the top
    // host last showed — re-read on every render, this host's own "first" turned silent the moment
    // "second" arrived (its own screen had become the handover by then).
    render(<Toaster />);
@@ -124,7 +124,7 @@ describe("Toaster — handing the toasts to another host", () => {
    // A dialog that notifies and closes in ONE handler (UploadDialog: "added to the queue" + close)
    // puts its toast into the very batch that hands the app host the top spot. Classifying it from
    // the store live at the takeover render counted it among the inherited, so it went in with no
-    // role into a region that was still `aria-live="off"` — announced by neither path (round 5).
+    // role into a region that was still `aria-live="off"` — announced by neither path.
    function Stage() {
      const [open, setOpen] = useState(true);
      return (
@@ -4,19 +4,19 @@
 // viewport at z-overlay paints over it, so a toast raised while that overlay is open is unreadable and
 // a click on it lands on the overlay's backdrop. The fix is a convention — the overlay mounts
 // `<Toaster overlay order={useOverlayOrder()} />` inside itself — and a convention stated in a comment
-// is one the next overlay forgets (deep-review round 3 found every ui/Modal and the onboarding wizard
+// is one the next overlay forgets (a review found every ui/Modal and the onboarding wizard
 // without one). So it is checked here: a component that renders a viewport-covering layer at
 // z-overlay or above mounts an overlay host, ranked by when the overlay opened.
 //
 // The class TOKENS are matched in any order, and the layer need not be `inset-0`: the first version
 // matched the exact string "fixed inset-0 z-overlay", so PlexPlayer's own root (`fixed left-0 top-0
 // z-overlay`) was never checked at all, and a new viewer copying it would have slipped past
-// (deep-review round 4).
+// (found by a later review).
 import { describe, it, expect } from "vitest";

 // BOTH patterns: `../**/*.tsx` from this directory does not match the siblings in `ui/` itself, so
 // ui/Modal.tsx — the overlay behind every dialog in the app — was never checked by this guard at all
-// (deep-review round 4).
+// (found by a later review).
 const sources = import.meta.glob(["../**/*.tsx", "./*.tsx"], {
  query: "?raw",
  import: "default",
@@ -3,11 +3,11 @@
 // The toast handover under React 18 StrictMode — its own file, because `lastTopIds` is module-level
 // state that leaks between tests.
 //
-// Round 5 made a taking-over host inherit "what the PREVIOUS top host last had on screen" by reading
+// One fix made a taking-over host inherit "what the PREVIOUS top host last had on screen" by reading
 // a module global in a `useMemo` and overwriting it in the same render pass. React re-invokes a
 // component body — and every `useMemo` factory in it — on a StrictMode double render, and the LAST
 // invocation is what commits: so the second pass re-read what the first had just written, the host
-// classified its OWN new toast as inherited, and the dev build behaved exactly like the code round 5
+// classified its OWN new toast as inherited, and the dev build behaved exactly like the code it
 // replaced. Measured in Chrome and in jsdom: role "status" without StrictMode, null with it.
 //
 // The named case is UploadDialog's — notify and close the overlay in ONE handler, so the toast is
@@ -22,7 +22,7 @@ import { SURFACE } from "./surfaces";
 // "Opened last", not "host mounted last": a host can remount while its overlay stays open —
 // PlayerModal's moves into the fullscreen stage and back — and ranked by mount time it came back on
 // top of a dialog opened over the player (the download dialog after leaving true fullscreen), so the
-// toasts were drawn under that dialog's scrim again (deep-review round 4). So the OVERLAY takes its
+// toasts were drawn under that dialog's scrim again. So the OVERLAY takes its
 // rank once, when it opens (`useOverlayOrder`, in overlayOrder.ts), and hands it to every host it renders.
 type Host = { id: string; order: number };
 const overlayHosts = createStore<readonly Host[]>([]);
@@ -70,16 +70,16 @@ export default function Toaster({
  // the takeover render. Reading the store live also silenced a toast RAISED in the same commit as
  // the takeover: a dialog that notifies and closes in one handler (UploadDialog) put its success
  // toast up in the very batch that handed the app host the top spot, so it was classified as
-  // inherited and announced by neither path (deep-review round 5). The top host records its screen
+  // inherited and announced by neither path. The top host records its screen
  // after each commit (the layout effect below); a host taking over reads it in a LATER render, after
  // that commit, so it always sees the outgoing host's last committed screen. (It was written during
-  // render, which made the React Compiler skip the Toaster — E32 — and let a discarded render
+  // render, which made the React Compiler skip the Toaster and let a discarded render
  // overwrite it.)
  //
  // Snapshotted into state when `isTop` flips, adjusted DURING render — not a `useMemo`, which React
  // may recompute (and the compiler is free to): a recompute while top would re-read this host's OWN
  // screen and silence its next toast. Under StrictMode the double invocation reads the same,
-  // unchanged global, so both passes agree (the round-6 failure was a render-phase WRITE racing it).
+  // unchanged global, so both passes agree (an earlier failure was a render-phase WRITE racing it).
  const [handover, setHandover] = useState<{ top: boolean; ids: ReadonlySet<number> }>({
    top: false,
    ids: EMPTY_IDS,
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E11 S4: a hover surface (tooltip, hover card) is a PASSIVE layer. It still sits on the layer stack —
+// A hover surface (tooltip, hover card) is a PASSIVE layer. It still sits on the layer stack —
 // Escape closes it before the surface beneath — but it does not own the keyboard, so the page-level
 // keys (`j`/`k`, `s`, the `g` chords) keep working while one happens to be open under a resting pointer.
 // Without the flag, hovering any hinted icon silently disabled keyboard navigation.
@@ -7,7 +7,7 @@ import { SURFACE } from "./surfaces";
 type Side = "top" | "bottom" | "left" | "right";

 /** Wrap any element to show a short glass hint caption on hover — by default only while the
- *  app-wide hints toggle (Settings → Appearance) is on. Anchored by the R8 popover platform, so it
+ *  app-wide hints toggle (Settings → Appearance) is on. Anchored by the popover platform, so it
 *  is never clipped by an overflow/stacking ancestor, flips when the preferred side runs out of room
 *  and slides back inside the viewport near an edge.
 *  `always` bypasses the hints toggle (for a label that IS the primary affordance, e.g. the slim nav
@@ -21,7 +21,7 @@ export default function UndoToolbar({
 }) {
  const { t } = useTranslation();
  // Only while a key actually DOES something: the sheet must not advertise Ctrl+Z on a page where
-  // nothing has been done yet and the handler returns immediately (E10 review round 1) — the same
+  // nothing has been done yet and the handler returns immediately — the same
  // honest-states gate PlexBrowse uses for its Backspace.
  useKeyScope("undo", canUndo || canRedo);

@@ -44,7 +44,7 @@ export interface VirtualGridProps<T> {
  /** Single-column layout only (`minCol == null`): the reading-width cap in px (null = fill). */
  colMaxWidth?: number | null;
  /** Multi-column layout only (`minCol != null`): the most columns to lay out, or null for as many
-   * as fit. This is the UPPER end of the viewport contract (E10 S4): an auto-fill grid always settles
+   * as fit. This is the UPPER end of the viewport contract: an auto-fill grid always settles
   * just above its minimum column, so on an ultrawide it answers "more room" with more, smaller
   * items forever — 3440px gave the feed 11 columns of 264px. With a cap the extra width goes into
   * the items instead (8 columns of ~370px), which is what a screen watched from a distance wants.
@@ -131,7 +131,7 @@ export default function VirtualGrid<T>({

  const rows = useMemo(() => chunk(items, colCount), [items, colCount]);

-  // The React Compiler (E27 S1) SKIPS memoising this whole component, and says so: TanStack
+  // The React Compiler SKIPS memoising this whole component, and says so: TanStack
  // Virtual's `useVirtualizer()` hands back functions that cannot be memoised without going stale,
  // so the compiler bails rather than risk a frozen viewport. That is the correct trade here — this
  // component's cost is the row window it mounts, not its render closures — and it is the ONE such
@@ -23,7 +23,7 @@ import {
 import { useLayer } from "../../lib/layerStack";
 import { pageStep, pageTarget, scrollParent } from "../../lib/listPaging";

-// The non-component half of the R8 popover platform: the `useAnchoredPopover` engine hook and the
+// The non-component half of the popover platform: the `useAnchoredPopover` engine hook and the
 // `closeThenAct` row helper. The surfaces it drives (PopoverSurface / PopoverItem) and the platform's
 // design notes live in AnchoredPopover.tsx; the split keeps that file components-only (Fast Refresh).

@@ -69,13 +69,13 @@ export interface AnchoredPopoverOptions {
  /** Close when a scroll container under the panel scrolls. Off by default, because a panel anchored
   *  to a real trigger element simply MOVES with it. Turn it on when the anchor is a fixed point
   *  rather than an element — a right-click menu, which otherwise stays parked over whatever row
-   *  scrolls under it while its actions still target the row you opened it on (E10 review round 1).
+   *  scrolls under it while its actions still target the row you opened it on.
   *  This is also what a native context menu does. */
  dismissOnScroll?: boolean;
  /** Menu/listbox only: ↑/↓ on the CLOSED trigger open the panel (the APG menu-button default, on).
   *  Turn it off where those keys already mean something on that trigger — a media player's row,
-   *  where ↑/↓ are the volume from anywhere, and a trigger that swallowed them killed it (E28 review
-   *  round 2, P2). Enter/Space/click still open it, and the arrows still rove once it is open. */
+   *  where ↑/↓ are the volume from anywhere, and a trigger that swallowed them killed it.
+   *  Enter/Space/click still open it, and the arrows still rove once it is open. */
  openOnArrowKey?: boolean;
 }

@@ -308,7 +308,7 @@ export function useAnchoredPopover(opts: AnchoredPopoverOptions) {
 /** A menu row whose action opens a dialog or moves the focus: close the menu FIRST and act once it
 *  has handed the focus back. floating-ui returns the focus to the trigger from a microtask queued as
 *  the menu unmounts, so an action run in the same batch had its focus taken back by it: measured in
- *  Chrome (E11 S2), a confirm dialog opened from the rail's menu landed its focus on Cancel and lost it
+ *  Chrome, a confirm dialog opened from the rail's menu landed its focus on Cancel and lost it
 *  7 ms later to the row BEHIND its scrim, where Enter activated the row. A fresh task runs after that
 *  microtask — and a dialog opened there captures the trigger, not a dying menu item, as the element
 *  to restore the focus to. Every such row goes through here (`ActionMenu`, "Manage views…"). */
@@ -2,7 +2,7 @@ import { createStore } from "../../lib/store";

 // Whether the Back-to-top button is up. It floats over the page's bottom-right corner, so PageScroller
 // reads this to keep the END of the page clear of it while it shows: a table's closing pager sat under
-// the button — "100" and "All" covered (E28 UAT, the channel manager on a short window).
+// the button — "100" and "All" covered (found in UAT, the channel manager on a short window).
 export const backToTopShown = createStore(false);

 export function useBackToTopShown(): boolean {
@@ -151,7 +151,7 @@ export function buttonClass({
    // The coarse-pointer target (DESIGN.md A3) lives HERE, not in the call sites — on every variant.
    // The link variants were once excluded as "a run of text inside a sentence", but none is: every
    // one stands on its own ("See all", "Forgot your passphrase?", a notification's action), and the
-    // E28 audit measured them at 16–20px on a tablet. Text inside a sentence is a plain <a>.
+    // an audit measured them at 16–20px on a tablet. Text inside a sentence is a plain <a>.
    "touch-target",
    fullWidth && "w-full",
  );
@@ -4,7 +4,7 @@ import { useEffect, useEffectEvent, useRef, useState, type RefObject } from "rea
 // through the shell; the onboarding wizard renders its own shell (a bare `ui/Overlay`, because it is
 // not a titled dialog with a header row) and used to have NONE of this — no trap, no initial focus,
 // no restore — so a keyboard user tabbed straight out of the first-login wizard into the app behind
-// the scrim, where the focus ring is invisible (deep-review round 5). A second hand-written copy is
+// the scrim, where the focus ring is invisible. A second hand-written copy is
 // how two shells drift apart; this is the shared one.

 // Tabbable elements inside the dialog, in DOM order. `:not([tabindex="-1"])` drops programmatically
@@ -19,7 +19,7 @@ export const focusablesIn = (root: HTMLElement): HTMLElement[] =>
 /** A form's first field — where a dialog opened to fill something in takes the focus: its first
 *  control, unless one is marked `data-initial-focus`. The mark is for a form with a control above
 *  its fields that is not one of them — the issue composer's Edit / Preview switch took the focus,
- *  and typing went nowhere (E32 review). */
+ *  and typing went nowhere. */
 export const firstFieldIn = (root: HTMLElement): HTMLElement | undefined => {
  const all = focusablesIn(root);
  return all.find((el) => el.hasAttribute("data-initial-focus")) ?? all[0];
@@ -58,7 +58,7 @@ export function trapTab(
 // ONE body scroll lock for every open dialog, counted. Each dialog used to save and restore
 // `overflow` on its own, so two unmounting out of order (the player's cleanup running before a Modal
 // opened above it) put back a stale "hidden". Invisible today — the app scrolls inside <main>, not
-// the document (measured, E11 S2) — but a count cannot be restored out of order, and PlayerModal's
+// the document (measured) — but a count cannot be restored out of order, and PlayerModal's
 // second copy of the lock is gone.
 let scrollLocks = 0;
 let unlockedOverflow = "";
@@ -100,7 +100,7 @@ export function useDialogChrome(
  // the cleanup hands focus back to the trigger, and the second mount then found the focus outside
  // the dialog and moved it to the first focusable — taking it off the `autoFocus` field a beat
  // after the browser had put it there. Measured in Chrome: the command palette opened with the
-  // focus on its ✕ in dev and on its search field in prod, from the same code (round 5).
+  // focus on its ✕ in dev and on its search field in prod, from the same code.
  const focusedRef = useRef<HTMLElement | null>(null);
  // Whether the dialog still holds the focus. Read at cleanup rather than computed there: by the
  // time a passive cleanup runs on a real unmount the panel is already out of the document and the
@@ -112,7 +112,7 @@ export function useDialogChrome(
  // stays reachable — so tabbing to that ✕ read as "the user left the dialog", and since removing
  // the toast fires no `focusin` to re-arm, closing the dialog afterwards stranded the focus on
  // <body> instead of handing it back to the trigger. Measured in jsdom, and a regression: before
-  // this hook existed the restore was unconditional (deep-review round 6).
+  // this hook existed the restore was unconditional.
  const heldRef = useRef(true);
  // Read at open, not a dependency: a caller's fresh function each render must not re-run the open.
  const pickInitial = useEffectEvent(() => initial?.() ?? null);
@@ -144,7 +144,7 @@ export function useDialogChrome(
      // it — and while it held the focus we read that as the user walking away. Then both closing
      // together left the focus on <body>: ours skipped the restore, and the child's own
      // `restoreTo` was already detached (passive cleanups run parent-first). Measured in jsdom,
-      // and it is the ordinary case — an error dialog raised over a dialog (round 6).
+      // and it is the ordinary case — an error dialog raised over a dialog.
      heldRef.current =
        (owner?.contains(active) ?? false) ||
        (active instanceof Element && active.closest('[aria-modal="true"]') != null);
@@ -171,7 +171,7 @@ export function useDialogChrome(
      // a layer closing with us is already disconnected while one that stays open is not. Without
      // this, a dialog closing on its own (an in-flight mutation's `onSuccess`) under a modal raised
      // above it pulled the focus out of that standing dialog onto our own trigger, behind its
-      // scrim — where Enter still activated it (deep-review round 7).
+      // scrim — where Enter still activated it.
      const owner = ownerAtOpen;
      const active = document.activeElement;
      const above = active instanceof Element ? active.closest('[aria-modal="true"]') : null;
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 audit: the Switch was a bare 36×20 track — under the 24 floor on a mouse, never the 44 touch box
+// Audit: the Switch was a bare 36×20 track — under the 24 floor on a mouse, never the 44 touch box
 // DESIGN.md A3 lists it under, and its name was optional, so the Plex-sync switch shipped nameless.
 import { afterEach, describe, expect, it, vi } from "vitest";
 import { cleanup, fireEvent, render, screen } from "@testing-library/react";
@@ -17,7 +17,7 @@ describe("columnsFor", () => {
  // The GUTTERS have to be in the count, not just in itemWidthFor. Every other assertion in this file
  // survives replacing `floor((w + GAP) / (minCol + GAP))` with the naive `floor(w / (minCol + GAP))`
  // — the fixed widths agree, the capped cases are cap-bound, and the two loops compare columnsFor
-  // against itself, so they cancel any shared change (E10 review round 1). These are the exact-fit
+  // against itself, so they cancel any shared change. These are the exact-fit
  // boundaries where the two formulas first disagree: an n-column row needs n*minCol + (n-1)*GAP, so
  // at exactly that width the naive form loses the last column and wastes a third of the row.
  it("counts the gutters, not one gap per column", () => {
@@ -3,7 +3,7 @@
 // the grid simply drops to the columns that fit).
 //
 // An auto-fill grid answers "more room" with more columns, never bigger ones, so it always settles
-// just above its minimum column width no matter how wide the screen is. Measured at 3440px (E10 S4),
+// just above its minimum column width no matter how wide the screen is. Measured at 3440px,
 // the feed laid out 11 columns of 264px — a diluted feed is as bad a failure of the viewport contract
 // as a broken one, so `maxCols` stops the multiplication and lets the leftover width flow into the
 // items instead. The cap can only bind ABOVE the width at which that many columns already fit, so the
@@ -1,7 +1,7 @@
 import { readdirSync, readFileSync } from "node:fs";
 import { describe, expect, it } from "vitest";

-// E28 review F2: a view that applies the column filters itself (`filterRows`) — DataTable, the
+// A view that applies the column filters itself (`filterRows`) — DataTable, the
 // playlist manager's own table, the channel manager's cards — can hide the column a filter lives in,
 // or have no header at all. Each renders the hidden-filter cue, so no filter narrows its rows unseen.
 describe("every view that filters rows by column", () => {
@@ -22,14 +22,14 @@ export const MEDIA_BADGE =
 *  Every picture box carries it: `mediaPictureBox` and the video card's own box. A box that holds a
 *  hover row (MEDIA_BAR) also sets `--bar-room`, the band the row needs: the narrow-picture rule
 *  below lifts the Play by it, and a box with no row (the feed's row and tile views, a Plex card
- *  without actions) lifts it by nothing (E31 review 2: it sat off-centre there). */
+ *  without actions) lifts it by nothing (it once sat off-centre there). */
 export const MEDIA_PICTURE_CONTAINER = "@container/pic has-[.media-bar]:[--bar-room:2.5rem]";

 /** The hover layer over the picture: a dim that carries the Play (mouse only, like the hover row).
 *  On a narrow picture the Play centres in the band ABOVE the hover row instead of on the whole
 *  picture: the picture's height follows the card's width, but the Play and the row grow with the
- *  text size, so at 1.3 a 242px thumbnail put the 63px Play 13px into the 37px row (E31 S5,
- *  measured). The cut is in rem, so it moves with the text size too: a picture under 14rem wide.
+ *  text size, so at 1.3 a 242px thumbnail put the 63px Play 13px into the 37px row
+ *  (measured). The cut is in rem, so it moves with the text size too: a picture under 14rem wide.
 *  Only where a row is there to clear (`--bar-room`, MEDIA_PICTURE_CONTAINER). */
 export const MEDIA_HOVER_LAYER =
  "reveal-hover absolute inset-0 flex items-center justify-center gap-2 bg-black/45 opacity-0 transition-opacity group-hover:opacity-100 group-focus-within:opacity-100 @max-[14rem]/pic:pb-(--bar-room)";
@@ -3,8 +3,8 @@ import { describe, expect, it } from "vitest";

 // A card skips its off-screen rendering (`cv-card` / `cv-poster`: content-visibility auto with a
 // reserved height), but not inside a carousel. A rail stretches every card to its tallest, and an
-// off-screen card's reserve WAS the tallest — Home's shelves measured 338px around 170–216px cards
-// (E28 S3h). The pair lives in two files and neither can be seen in jsdom, so it is held here.
+// off-screen card's reserve WAS the tallest — Home's shelves measured 338px around 170–216px cards.
+// The pair lives in two files and neither can be seen in jsdom, so it is held here.
 const read = (file: string) => readFileSync(new URL(file, import.meta.url), "utf8");

 describe("a carousel turns the cards' content-visibility off", () => {
@@ -9,7 +9,7 @@ const sources = () =>

 /** Checkable menu rows a file renders: a literal `role="menuitemradio"`, and a role CHOSEN by an
 *  expression that can come out checkable — FilterSelect's and ActionMenu's rows, every filter select
- *  and checkable action menu in the app, which a literal-only count skipped (round 3). */
+ *  and checkable action menu in the app, which a literal-only count skipped. */
 function checkableRoles(src: string): number {
  const literal = (src.match(/role="menuitem(?:radio|checkbox)"/g) ?? []).length;
  const chosen = [...src.matchAll(/role=\{([^{}]*)\}/g)].filter(([, e]) =>
@@ -18,7 +18,7 @@ function checkableRoles(src: string): number {
  return literal + chosen;
 }

-// E28 review L2: `ui/MenuCheck` is the ONE drawn check box of a checkable row (S3e3b replaced seven
+// `ui/MenuCheck` is the ONE drawn check box of a checkable row (S3e3b replaced seven
 // hand-drawn copies), yet FilterSelect and Combobox still painted their own — a 14px box beside the
 // menus' 16px one. The filled-box paint is the tell: it may live in MenuCheck only.
 describe("the drawn check box", () => {
@@ -30,7 +30,7 @@ describe("the drawn check box", () => {
    expect(offenders).toEqual([]);
  });

-  // Round 2 (P6): the player's track rows marked the choice only by an accent label — which the
+  // The player's track rows marked the choice only by an accent label — which the
  // highlight's white text wiped — and four more menus ticked it their own way (a trailing tick, a
  // dot, a filled row). A file with a checkable menu row draws MenuCheck, and no tick of its own.
  it("marks every checkable menu row, with no tick of its own beside it", () => {
@@ -48,7 +48,7 @@ const PLAIN: Record<string, [number, string]> = {
  "ui/CommandPalette.tsx": [1, "a command row"],
  "ui/NavRail.tsx": [
    2,
-    "the action rows (About, Sign out) and the group row that opens its panel; a page row is a link (E31)",
+    "the action rows (About, Sign out) and the group row that opens its panel; a page row is a link",
  ],
  "ui/PanelGroup.tsx": [2, "a group's collapse toggle and its label"],
  "ui/SidePanel.tsx": [2, "panel controls"],
@@ -1,7 +1,7 @@
 import { startTransition, useEffect, useState } from "react";

 // A long page of table rows mounts in steps: the first FIRST_ROWS at once, then STEP_ROWS more per
-// background transition until all are in (E31 S4d). The channel manager's "All" page (~400 rows) took
+// background transition until all are in. The channel manager's "All" page (~400 rows) took
 // 0.7–0.9 s of one blocked frame to mount on prod, and every sort and every keystroke in its search
 // re-rendered all of them (300–400 ms per key). No single cell is to blame: measured column by column,
 // each costs its share, and all of them together cost ~9× an empty row. The first rows fill more than
@@ -21,7 +21,7 @@ const STEP_ROWS = 40;
 //
 // `through` is a row that must be in the DOM in THIS render: the row a page turn was made to show
 // (DataTable's `reveal`). Its page is a new view, and the caller focuses that row right after the
-// commit — at index 40 of a 50-row page it was not mounted yet, and focus fell to <body> (E31 S5).
+// commit — at index 40 of a 50-row page it was not mounted yet, and focus fell to <body>.
 export function useProgressiveRows<T>(rows: T[], view: string, through = -1): T[] {
  const [seen, setSeen] = useState(view);
  const [limit, setLimit] = useState(FIRST_ROWS);
@@ -1,4 +1,4 @@
-// A card's hover chrome (E28 C1): shown when the card is hovered or holds keyboard focus on a mouse,
+// A card's hover chrome: shown when the card is hovered or holds keyboard focus on a mouse,
 // removed on touch (`.reveal-hover`, index.css), where the card offers the same actions through its
 // "…" beside the title. One string, so the video card and the Plex cards reveal the same way.
 export const HOVER_REVEAL =
@@ -1,4 +1,4 @@
-// E31 (the owner's pick): the channel page's tall header folds to one row once the grid scrolls. The header
+// The owner's pick: the channel page's tall header folds to one row once the grid scrolls. The header
 // is fixed, so folding gives the scroller its height back — these are the rules that keep that from
 // flickering on a page barely taller than the window.
 import { describe, expect, it } from "vitest";
@@ -1,7 +1,7 @@
 import { useEffect, useRef, useState, type RefObject } from "react";
 import { CONTENT_ID } from "./landmarks";

-// A tall page header that folds to one row once the page's content scrolls (E31, the owner's pick for the
+// A tall page header that folds to one row once the page's content scrolls (the owner's pick for the
 // channel page: the banner and the identity block took ~40% of a 133% screen before the first video).
 //
 // The header is FIXED (it sits in the shell's toolbar band, outside the scroller), so folding it makes the
@@ -23,7 +23,7 @@ import { SURFACE } from "./surfaces";
 // Why it lives here and not in DataTable: DataTable owns its filter/sort/page state internally,
 // while the playlist manager's body is a drag-reorderable list whose order is the persisted
 // server-side one. Two owners of the same ordering fight; composing these pieces instead lets the
-// playlist keep its stored order and still get real column headers. Same move as E4 S3's
+// playlist keep its stored order and still get real column headers. Same move as
 // ChannelLayoutGrid, which builds a card view straight from the `Column<T>` render closures.

 // The multi-select filter's option list, capped and scrollable. Its own component because it owns a
@@ -46,7 +46,7 @@ function FilterOptionList({ children }: { children: ReactNode }) {
 // so the caret jumped to the end and IME composition (accents, any CJK input) broke mid-word. The
 // `autoFocus` masked it well enough that it read as "the filter is just twitchy".
 //
-// The funnel and its panel live in ONE component because the R8 popover platform anchors the panel
+// The funnel and its panel live in ONE component because the popover platform anchors the panel
 // to its trigger. That also retires this file's three hand-rolled focus chores: capturing the
 // trigger element at open time (a `ref` scoped to the open column nulled out before focus could go
 // back), the restore-focus effect that stopped Escape stranding focus on <body>, and the useDismiss
@@ -14,7 +14,7 @@ import { type ChipSize, type ChipVariant, chipClass } from "./chipStyles";

 // Tailwind generates CSS only for the class names it FINDS in the source, token by token. A class
 // assembled at runtime (`${ACT}hover:text-fg`) is never found whole, so it ships with no rule at all:
-// every hover colour of Button, IconButton and Chip was dead that way, found in the E28 audit by
+// every hover colour of Button, IconButton and Chip was dead that way, found in an audit by
 // reading the stylesheet (0 rules for `not-aria-disabled`). So: every class a primitive can emit has
 // to stand in its source file whole, between the delimiters the scanner splits on.
 const tokens = (file: string) =>
@@ -10,7 +10,7 @@ import { matchesShortcut, useKeyScope } from "../../lib/keymap";
 import { pointOnSelection } from "../../lib/textSelection";
 import { useAnchoredPopover } from "./anchoredPopoverHooks";

-// E10 S2 — the right-click context menu's INPUT half: the gestures that open it and the row they
+// The right-click context menu's INPUT half: the gestures that open it and the row they
 // belong to. The surface itself is `<ContextMenu>` (ContextMenu.tsx); they are two files only so
 // that the component file exports a component and nothing else (react-refresh).
 //
@@ -68,8 +68,8 @@ export interface ContextMenuController<T> {
  anchor: Element | null;
  /** Spread on the element that should answer a right-click — the whole row, not a handle. */
  triggerProps: (payload: T) => Record<string, unknown>;
-  /** Spread on the row's visible "…" button (E28 C1: nothing is reachable only by a gesture). A
-   *  click opens the SAME menu under the button; a second click on it while its menu is open closes
+  /** Spread on the row's visible "…" button (DESIGN.md C1: nothing is reachable only by a gesture).
+   *  A click opens the SAME menu under the button; a second click on it while its menu is open closes
   *  it, like any menu button. Pair it with `aria-expanded` from the caller, which knows its row. */
  buttonProps: (payload: T) => Record<string, unknown>;
  /** Opens the menu at a viewport point for `payload`, anchored to `contextElement` — for an input
@@ -148,7 +148,7 @@ export function useContextMenu<T>({
  // never produce that click (a pan ends in pointercancel, the finger lifts over the portalled menu,
  // Android Chrome suppresses the trailing click after its own contextmenu). The flag then stayed
  // armed, and since it is one ref per LIST, the next genuine tap on ANY row of that list was eaten
-  // and the row looked dead until tapped twice (E10 review round 1). So a fresh press also disarms
+  // and the row looked dead until tapped twice. So a fresh press also disarms
  // it: whatever the last gesture left behind cannot outlive the start of the next one.
  const swallowClick = useRef(false);
  const cancelPress = useCallback(() => {
@@ -1,6 +1,6 @@
 // @vitest-environment jsdom
 //
-// E28 audit: an in-place editor closed by a REAL Enter re-opened at once. The keydown commits and hands
+// Audit: an in-place editor closed by a REAL Enter re-opened at once. The keydown commits and hands
 // the focus back to the trigger button; the same keystroke's default action (the keypress activation)
 // then lands on that button and clicks it. A synthetic keydown has no default action, so the first
 // test passed while the live rename was broken. Arming now cancels the key's default action.