feat(admin): surface the sign-ups that are stuck outside, and let an admin finish letting them in (E9.5 S5)

Approve an address BEFORE the person registers and their sign-up produces no queue row at all —
`_decide` only takes a still-pending invite — so the Access-requests screen sat empty, the tab badge
read zero, and the only working move was re-typing the address into the whitelist box. Nobody would
guess that, and on 2026-09-03 nobody did.

This is roadmap UX-25, where two earlier attempts were reverted after review. Both moved AUTHORITY:
auto-activating on registration turns a standing whitelist entry into a bearer grant for whoever
registers the address first, and reopening the approved invite to `pending` lets an anonymous sign-up
REVOKE a standing approval. What was actually missing was visibility, so this changes nothing about
what an unauthenticated caller can cause — the admin could already do exactly this by re-entering the
address; they just had no way to know they had to.

The section lists every account that cannot sign in yet and says WHICH gate is shut, because the two
dead ends look the same from the admin's chair: waiting for activation, or waiting on a confirmation
mail. Naming the unconfirmed ones is not decoration — on a no-SMTP instance that is the difference
between the owner and whoever registered the address first (SEC-15), and an admin about to let
someone in should see it. A person the queue above already covers is deliberately NOT listed here;
two buttons for one person is worse than one.

Activate is gated on `is_allowed` — the same function the sign-in paths consult, in core, so the
button that appears and the badge that counts cannot disagree. Without that gate it would
manufacture an inconsistent account: password sign-in only checks `is_active` and would let them in,
while their Google sign-in would be turned away and silently reopened as a fresh access request.
Letting in someone with no approved invite stays the whitelist box's job, which says so out loud.

The badge counts activations SEPARATELY from pending invites rather than widening an existing
field's meaning, and only counts what genuinely waits on the admin: an unconfirmed account waits on
its owner, so it is listed with a re-send but does not nag. The admin-side resend says plainly what
happened, including refusing when the instance has no working SMTP — mail is fail-soft everywhere
else, and that silence is exactly what left the 0.78.5 sign-ups with nothing to fall back on.

`send_verification_link` is now one function with three callers (registration, the public resend,
this one), so the link's shape and TTL cannot drift by who triggered it.

UAT on localdev, both languages: the UX-25 shape reproduced end to end (approve the address, then
register), the badge counting it, Activate letting them in and dropping the badge, the row losing
its Activate but keeping the re-send, a fresh token issued by the re-send, and both actions in the
audit log. Test data removed.

backend/app/routes/admin.py

@@ -6,7 +6,7 @@ from datetime import datetime, timezone
 
 from fastapi import APIRouter, BackgroundTasks, Depends, HTTPException
 from pydantic import BaseModel
-from sqlalchemy import select
+from sqlalchemy import or_, select
 from sqlalchemy.orm import Session
 
 from app import audit
@@ -14,8 +14,10 @@ from app import email as email_mod
 from app.audit import AuditAction
 from app.auth import (
     admin_user,
+    can_activate,
     count_admins,
     get_or_create_demo_user,
+    is_allowed,
     operator_contact,
     purge_user,
 )
@@ -27,6 +29,8 @@ from app.models import (
     Invite,
     User,
 )
+from app.routes._common import get_or_404
+from app.routes.auth import send_verification_link
 from app.utils import canonical_email, iso, valid_email
 
 router = APIRouter(prefix="/api/admin", tags=["admin"])
@@ -414,6 +418,133 @@ def admin_delete_user(
 # --- Demo account: email whitelist + state reset -------------------------------------
 
 
+# --- Stuck sign-ups: the admin lifeline (E9.5 S5, roadmap UX-25) ---------------------------------
+
+
+def _pending_access_rows(db: Session) -> list[User]:
+    """Accounts that have signed up but still cannot sign in, newest first.
+
+    Two different dead ends share this list, because from the admin's chair they look the same — a
+    person who registered and is still outside:
+      · `is_active=False` — approved BEFORE they signed up, so nothing ever landed in the
+        Access-requests queue and its Approve button doesn't apply (`_decide` only takes a pending
+        row). This is roadmap UX-25, and it is why the list exists.
+      · `email_verified=False` — the confirmation mail hasn't been acted on. Not the admin's to
+        decide, but theirs to help with: on 2026-09-03 the first mail to a fresh Gmail landed in
+        Spam, so "send it again" is a real move.
+    A suspended account is deliberately excluded: it cannot sign in either, but that is a decision
+    already made, not a person stuck waiting on one.
+    """
+    return list(
+        db.execute(
+            select(User)
+            .where(
+                User.is_demo.is_(False),
+                User.is_suspended.is_(False),
+                or_(User.is_active.is_(False), User.email_verified.is_(False)),
+            )
+            .order_by(User.created_at.desc())
+        )
+        .scalars()
+        .all()
+    )
+
+
+@router.get("/pending-access")
+def list_pending_access(_: User = Depends(admin_user), db: Session = Depends(get_db)) -> list[dict]:
+    """The sign-ups that are still outside, with which gate is shut for each. `can_activate` cannot
+    be derived on the client — it needs the invite/allow-list lookup — so it is answered here.
+
+    One `is_allowed` query per row rather than a batched re-implementation of it: the row count is
+    the admin's own backlog of stuck sign-ups (single digits on any real instance), and a second,
+    subtly different copy of the "may this address sign in" rule is exactly the drift this codebase
+    keeps getting bitten by."""
+    rows = [(u, can_activate(db, u)) for u in _pending_access_rows(db)]
+    # An account that is inactive, confirmed and merely waiting on a still-pending invite is ALREADY
+    # an actionable row in the Access-requests queue above, with Approve on it. Listing it here too
+    # would make the admin choose between two buttons for one person.
+    return [
+        _serialize_user(u) | {"can_activate": ca} for u, ca in rows if ca or not u.email_verified
+    ]
+
+
+@router.post("/users/{user_id}/activate")
+def activate_user(
+    user_id: int,
+    admin: User = Depends(admin_user),
+    db: Session = Depends(get_db),
+) -> dict:
+    """Let in a registrant whose address was approved before they signed up (roadmap UX-25).
+
+    Admin-side ONLY, and that is the whole design. Two earlier attempts at this were reverted after
+    review, and both moved authority rather than visibility: auto-activating on registration turns a
+    standing whitelist entry into a bearer grant for whoever registers the address first, and
+    reopening the approved invite to `pending` lets an anonymous sign-up REVOKE a standing approval.
+    This endpoint changes nothing about what an unauthenticated caller can cause — the admin could
+    already do exactly this by re-entering the address in the whitelist box (`add_invite` →
+    `_activate_user_by_email`). What was missing was any way to know they had to."""
+    target = get_or_404(db, User, user_id, detail="No such user")
+    if target.is_demo:
+        raise HTTPException(status_code=400, detail="The demo account can't be activated.")
+    if target.is_active:
+        return _serialize_user(target)  # idempotent: a double click is not an error
+    if not is_allowed(db, target.email):
+        raise HTTPException(
+            status_code=409,
+            detail={"code": "activate", "reason": "not_allowed", "email": target.email},
+        )
+    audit.record(
+        db,
+        admin.id,
+        AuditAction.USER_ACTIVATE,
+        target_type="user",
+        target_id=target.id,
+        summary=target.email,
+        before={"is_active": False},
+        after={"is_active": True},
+    )
+    target.is_active = True
+    db.commit()
+    return _serialize_user(target)
+
+
+@router.post("/users/{user_id}/resend-verification")
+def resend_user_verification(
+    user_id: int,
+    admin: User = Depends(admin_user),
+    db: Session = Depends(get_db),
+) -> dict:
+    """Re-send an account's confirmation link from the admin side.
+
+    The public /auth/verify/resend answers uniformly and reveals nothing, which is right for an
+    anonymous caller but leaves the admin unable to tell a delivered mail from a dropped one. Here
+    the caller is already authenticated as an admin, so this one says plainly what happened.
+
+    Not rate-limited: an admin is not a threat model to enumerate against, and the deliberate silence
+    of the public endpoint is exactly what this exists to escape."""
+    target = get_or_404(db, User, user_id, detail="No such user")
+    if target.is_demo or target.email_verified:
+        raise HTTPException(
+            status_code=409,
+            detail={"code": "resend", "reason": "not_applicable", "email": target.email},
+        )
+    if not email_mod.email_enabled(db):
+        # Mail is fail-soft everywhere else, which is what left the 0.78.5 sign-ups with nothing to
+        # fall back on. An admin pressing a button deserves the failure, not a success toast.
+        raise HTTPException(status_code=409, detail={"code": "resend", "reason": "email_off"})
+    send_verification_link(db, target)
+    audit.record(
+        db,
+        admin.id,
+        AuditAction.USER_VERIFY_RESEND,
+        target_type="user",
+        target_id=target.id,
+        summary=target.email,
+    )
+    db.commit()
+    return {"ok": True, "email": target.email}
+
+
 def _serialize_demo(row: DemoWhitelist) -> dict:
     return {
         "id": row.id,