ScholarFlow v2.4.0 - Enhanced PDF Extraction & i18n Optimization

Key improvements in this release:

Smarter PDF Metadata Extraction
- Increased buffer scan from 65KB to 256KB for comprehensive PDF metadata
- Added support for /Subject, /ModDate, /Keywords fields
- Prioritizes PDF metadata with automatic fallback to filename if metadata is generic
- Intelligently selects longer, more meaningful title between PDF metadata and filename
Example: "Document" (PDF) → "Machine Learning 2024.pdf" (filename)

Full i18n (Internationalization) Support
- Complete localization of "Local Document / PDF" label across 5 languages
(Vietnamese, English, Chinese, Japanese, Russian)
- Fixed citation formatting duplicates for local document labels
- Accessed-date formatting fully supports all languages

Bug Fixes
- Fixed ReferenceError: currentAppLanguage is not defined
- Optimized Vietnamese PDF title parsing: prevents over-capitalization of diacritics
Example: "HướNg DẫN ChấM..." → displays correctly

Enhanced Source Extraction
- Full support for extracting PDFs from local files (file://)
- PDF parsing from in-browser viewers
- Automatic fallback when direct PDF fetch fails

Version: 2.4.0
Channel: Stable