8baa07db84
pymupdf.layout.activate() causes pymupdf4llm.to_markdown() to ignore the page_chunks=True option, returning a single string instead of list[dict]. This broke per-page chunking needed for semantic search indexing. See: https://github.com/pymupdf/pymupdf4llm/issues/323 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>