refactor: R12 S3 — invert scheduler->routes, publish cross-module privates

Layering cleanup, no behaviour change (ruff clean backend+scripts, pytest 193):

- Invert the scheduler->routes dependency: reset_demo_state (+ its demo-seed
  helpers) move from routes/admin.py to a new service module app/demo.py.
  Both routes/admin.py and scheduler.py now import it from app.demo, so the
  scheduler no longer reaches into the routes layer — its lazy import (the
  routes<->scheduler cycle workaround) is gone, hoisted to a top-level import.

- Publish the private symbols that were imported across module boundaries (the
  underscore convention was lying): feed._serialize -> serialize,
  feed._to_tsquery_str -> to_tsquery_str, videos._insert_stubs -> insert_stubs,
  videos._stub_from_playlist_item -> stub_from_playlist_item,
  worker._uploader_url -> uploader_url, plex._shared._item_or_404 ->
  item_or_404. Scoped renames — the 5 other route modules keep their own truly
  private _serialize helpers.

- _release_asset moves from routes/downloads.py to app/downloads/service.py as
  the public release_asset (asset bookkeeping is a service concern); the routes
  call it via service.release_asset.

backend/app/plex/people.py

@@ -74,7 +74,7 @@ def enrich_row_people(db: Session, row, cast: list[dict]) -> bool:
         # items in the same batch: SessionLocal is autoflush=False, so the ORM query above can't see a
         # same-batch pending insert, and this path also races the concurrent info-page enrichment. The
         # Core execute runs immediately (not deferred to the caller's commit), so a later row in the
-        # batch sees the row and skips it. Mirrors app/sync/videos.py:_insert_stubs.
+        # batch sees the row and skips it. Mirrors app/sync/videos.py:insert_stubs.
         db.execute(
             pg_insert(PlexPerson)
             .values([{"name": nm, "photo_thumb": th} for nm, th in missing.items()])