Fixed
- Re-scanning no longer wipes API/manual language matches back to "heuristic". The scan upsert overwrote native_language + language_source with the freshly-scanned (heuristic) values, so a folder rescan or the watcher re-seeing a file downgraded every TMDB (api) or manual/tmdb-manual match to heuristic. The upsert now preserves an authoritative source/native and only refreshes heuristic/unset rows.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.94 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.94-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.94-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.94-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.