- Changed LLM model from llama3.2:1b to gpt-oss:20b-cloud in llm_engine.py files for page_rag and toc_rag.
- Added JSON export functionality for tree structures in indexer.py to support frontend visualization.
- Implemented retry logic for LLM API calls in toc_utils.py to handle 429 errors with exponential backoff.
- Enhanced title appearance checks in tree_builder.py with fast text-based checks to reduce LLM calls.
- Introduced a tree viewer modal in index.html and script.js for visualizing document structures.
- Added RAG mode toggle switches in the UI to allow users to select which indexing modes to use.
- Improved CSS styles for the tree viewer and RAG mode toggles for better user experience.
Add Table of Contents (TOC) RAG system with new modules, database enhancements, and UI integration.
