Implements automatic cleanup of completed, errored, or cancelled batches after 5 minutes to prevent stale state. Adds stuck task detection for 'searching' and 'post_processing' states, forcing failure or completion if tasks exceed time thresholds. Updates batch completion logic to track completion timestamps and handle orphaned tasks more robustly. Also improves wishlist auto-processing timer management.
Adds support for sending specific track IDs from the frontend to the backend when starting wishlist missing downloads. This ensures only the tracks currently visible to the user are processed, preventing race conditions if the wishlist changes between modal open and analysis start. Category filtering remains for backward compatibility.
ConfigManager now stores configuration in a SQLite database instead of a JSON file. The class supports migration from existing config.json files and falls back to defaults if no configuration is found. This change improves reliability and centralizes configuration management.
When an artist is added to the watchlist, their image is now fetched from Spotify and cached in the database immediately if available. This improves user experience by ensuring artist images are present without requiring a separate fetch.
Added logic to both automatic and manual wishlist processing to check for and remove tracks that already exist in the library before initiating downloads. This prevents unnecessary bandwidth usage and keeps the wishlist up to date.
Improves handling of concurrent wishlist and watchlist operations by moving scheduling outside the lock and using flags to avoid deadlocks. This change ensures that scheduling the next wishlist processing does not occur while holding the lock, reducing risk of deadlock and improving code clarity.
