[Feature] Quality Grading System for Media Assets

Summary

Implement a quality grading system for media assets that provides users with visibility into the quality of their library content.

Implementation Status

• [x] Database schema - Added quality columns to download_assets, media_items, and episodes tables
• [x] Quality Analyzer Service (app/services/quality_analyzer.py) - Multi-factor analysis including:
• Metadata scoring (40%): Resolution, HDR, source type, audio codec
• ML-based perceptual quality (30%): BRISQUE + NIQE metrics via PyIQA
• Bitrate efficiency (20%): Bitrate per pixel analysis
• Manual override support (10%): User adjustment capability
• [x] Integration with batch optimizer - Quality analysis runs after successful optimization
• [x] Branded tier naming: S (LOCKED ON), A (STRONG BEAM), B (TRACKING), C (WEAK SIGNAL), D (STATIC)

Remaining Work

• [ ] Add quality grade display in tvOS app UI
• [ ] Add quality grade display in web UI (movie/episode pages)
• [ ] Add API endpoint to manually override quality grades
• [ ] Add bulk quality analysis for existing library content
• [ ] Consider VMAF scoring for more accurate quality assessment

Technical Notes

• All processing is fully local (no external API calls after initial model download)
• Analysis takes ~15 seconds per file on M4 Max
• PyIQA models cached at ~/.cache/torch/hub/pyiqa/

Files Changed

• app/services/quality_analyzer.py (new)
• app/services/optimization_batch.py (modified)
• Database schema updated (quality columns added)