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.

frontend/src/components/ui/PageScroller.tsx

@@ -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.