Build release images without the layer cache

A submission failed with HTTP 403 against a freshly published image. The
cause was not Alpine, which carries the current yt-dlp: it was the apk layer
being reused from an older build, so the image shipped a yt-dlp two months
behind while /healthz reported today's tag.

"A rebuild is the update" only holds if the layer is genuinely re-run, so
the image target now passes --pull --no-cache. Slower, and the only thing
that makes the release honest about what is inside it.

docs/deployment.md

@@ -269,7 +269,7 @@ bounded by the two conversion slots.
 | Uploads fail near 50 MB | The reverse proxy's body limit, not the app's |
 | Invite links are relative | `PUBLIC_URL` unset |
 | Everything 500s after a restore | `-wal`/`-shm` sidecars from the replaced database were left in place |
-| Submissions all fail at download | yt-dlp is stale, or YouTube is refusing this server's IP; rebuild and publish the image first. `docker compose logs app \| grep '"stage":"download"'` shows yt-dlp's own stderr under `detail` |
+| Submissions all fail at download | yt-dlp is stale, or YouTube is refusing this server's IP. `docker compose logs app \| grep '"stage":"download"'` shows yt-dlp's own stderr under `detail`. A plain `HTTP Error 403` is the stale case — rebuild with `make image`, which forces `--no-cache` so the `apk add` layer is genuinely re-run. A `docker build` without it can ship a months-old yt-dlp from a cached layer |
 | A submitter reports a *virhekoodi* | `docker compose logs app \| grep <code>` — one line, with the stage, the submission id, the URL and the tool's stderr |
 | `/admin` returns 404 while logged in | That account has no `is_admin`. Set it in the database; nothing in the UI grants it |
 | Setting `ADMIN_PASSWORD` again changes nothing | Seeding only fires on an empty `users` table. Reset the hash in the database instead |