fix(audit-g1): six one-liner drift fixes from 2026-06-02 audit

- BACKFILL_TIMEOUT_SECONDS 1800→1170: keep the subprocess timeout
  30s below Celery's hard time_limit=1200 so SIGKILL doesn't beat
  TimeoutExpired (matched the tick 870s/900s rationale). Backfill
  runs that hit the cap let the next tick continue via the archive.

- recover_interrupted_tasks orphan UPDATE now stamps finished_at;
  without it cleanup_old_tasks' WHERE finished_at<cutoff never
  reaped orphan-swept rows. recover_stalled_task_runs also now sets
  duration_ms (matches celery_signals.finalize's millisecond math).

- ExtensionService.quick_add_source arms NEW_SOURCE_BACKFILL_RUNS=3
  on Source creation, mirroring SourceService.create. Without it,
  Firefox quick-add on a creator with >20 unsynced posts walked the
  full feed until subprocess timeout. Renamed the constant from
  _NEW_SOURCE_BACKFILL_RUNS so it can be imported cross-module.

- gallery_dl.verify() accepts TIER_LIMITED as auth-success alongside
  NO_NEW_CONTENT — the download path (line 712) already does, and
  TIER_LIMITED proves auth reached the post and was told it was
  tier-gated. Verify endpoint previously showed red on this and
  prompted operators to rotate working cookies.

- prune_unused_tags now runs a single DELETE with the NOT-IN
  predicate find_unused_tags uses, instead of SELECT-ids →
  DELETE-WHERE-IN. Removes the psycopg 65535-param cliff that
  would have surfaced on a tag explosion (>65k unused tags).

- credentials.upload() reflects the returned record into the store
  cache (`.set(platform, rec)`) instead of evicting it; previously
  the card briefly rendered "no credential" between upload and
  loadAll().

backend/app/services/cleanup_service.py

@@ -354,18 +354,35 @@ def prune_unused_tags(session: Session, *, dry_run: bool = False) -> dict:
     Returns:
       dry_run=True:  {"count": N, "sample_names": [first 50]}
       dry_run=False: {"deleted": N, "sample_names": [first 50]}
+
+    Implementation note: the previous SELECT-ids → DELETE-WHERE-IN
+    pattern was vulnerable to the psycopg 65535-parameter ceiling on
+    libraries with tag explosions. The live delete now runs a single
+    DELETE with the same NOT-IN predicate find_unused_tags uses, so
+    the row count scales without binding every id as a parameter.
+    Audit 2026-06-02.
     """
-    unused = find_unused_tags(session)
-    sample = [t.name for t in unused[:50]]
+    sample_rows = find_unused_tags(session, limit=50)
+    sample = [t.name for t in sample_rows]
+    used_via_image_tag = select(image_tag.c.tag_id).distinct()
+    used_via_series = select(SeriesPage.series_tag_id).where(
+        SeriesPage.series_tag_id.is_not(None)
+    ).distinct()
     if dry_run:
-        return {"count": len(unused), "sample_names": sample}
-    ids = [t.id for t in unused]
-    if ids:
-        session.execute(
-            Tag.__table__.delete().where(Tag.id.in_(ids))
-        )
-        session.commit()
-    return {"deleted": len(ids), "sample_names": sample}
+        count = session.execute(
+            select(func.count())
+            .select_from(Tag)
+            .where(Tag.id.not_in(used_via_image_tag))
+            .where(Tag.id.not_in(used_via_series))
+        ).scalar_one()
+        return {"count": count, "sample_names": sample}
+    result = session.execute(
+        Tag.__table__.delete()
+        .where(Tag.id.not_in(used_via_image_tag))
+        .where(Tag.id.not_in(used_via_series))
+    )
+    session.commit()
+    return {"deleted": result.rowcount or 0, "sample_names": sample}
 
 
 # Legacy tags FC no longer uses, in two shapes: