When _try_direct_stream() succeeded, it streamed chunks directly via
requests.get() without firing any progress events. This caused the WebSocket
frontend to only see 'started' (0%) and 'completed' (100%) — no incremental
updates.
Add per-chunk progress events to _try_direct_stream that mirror yt-dlp's
hook format (downloaded_bytes, total_bytes, speed, eta, status), emitted at
~1% intervals. These flow through the existing download_progress handler chain
(SeriesApp → anime_service → progress_service → WebSocket → frontend).
VOE provider returns HLS streams (.m3u8) which previously used yt-dlp's
native HLS downloader. That downloader does not report downloaded_bytes/
total_bytes in progress hooks, causing only 0% and 100% to ever show on
the queue page. Explicitly set 'downloader: ffmpeg' so progress updates
with actual byte counts are broadcast, enabling intermediate percentages
(25%, 50%, 75%, etc.) to display correctly.
- Add nfo_scan_after_rescan config option (default: true)
- Implement year caching in AniworldLoader and EnhancedAniWorldLoader
- Make get_year abstract method in base provider
- Run NFO validation/creation after scheduled rescan completes
- Add _YearDict cache to avoid re-extracting year from HTML
- Move src/core/ → src/server/
- Split SerieList.py (531 lines) and series.py (414 lines) into src/server/database/
- Add database/models.py for SQLAlchemy models
- Update all test imports to reflect new structure
- Remove deprecated test files (test_serie_class.py, test_serie_folder_with_year.py)
