Changes since version 3.2

We have implemented several major architectural improvements, new features, and security enhancements since v3.2:

1. New AI Providers & Local Models (v3.5 - v4.1):
2. Added support for multiple LLM providers, including DeepSeek and local models via Ollama.
3. Introduced user-configurable custom prompt templates.
4. Advanced Features (v3.5 - v4.2):
5. Implemented Batch Playlist Processing to summarize entire playlists.
6. Added interactive AI features: Generate Quiz based on transcript and Entity Extraction.
7. Support for YouTube Chapters and "Map-Reduce" chunking for very long videos.
8. Localization (v4.2):
9. Fully localized UI (English, Serbian, German, Spanish).
10. UX & Resilience Improvements:
11. Added a heartbeat timer during long AI analyses to prevent timeout issues.
12. Increased LLM timeout handling and implemented fallback mechanisms for API rate limits (429 errors).
13. Improved Markdown rendering contrast for timestamps on dark backgrounds.
14. Security & Architecture (v4.4.0):
15. Modularized prompt logic, summary rendering, and the transcript fetcher.
16. Resolved CodeQL security warnings regarding URL sanitization and multi-character regex matching.