fix(rag,extractors): resolve markdown table corruption and Content-Type parsing across document extractors

[vjymisal0]

Summary

Fixes multiple table formatting and header parsing issues across RAG document extractors (#41992, #41990, #41986, #41955):

1. Notion Extractor (Notion extractor corrupts Markdown tables for cells with mixed formatting or empty data cells #41992): Concatenate all rich-text segments within each Notion table cell into a single column string, preserve empty cells ([]) without column shifting, and escape pipe characters.
2. Word Extractor (Word extractor produces broken Markdown tables when a cell contains a pipe character #41990): Escape pipe characters (|) in Word table cells to prevent table delimiter breakage in generated Markdown tables.
3. Markdown Extractor (Markdown extractor corrupts code blocks containing angle brackets (<...> sequences are silently deleted) #41986): Preserve angle brackets (<...>) in fenced code blocks while cleaning HTML tags outside code blocks.
4. Extract Processor (load_from_url misroutes files when Content-Type has parameters (e.g. "application/pdf; charset=binary") #41955): Strip Content-Type parameters (e.g. ; charset=binary) before deriving file extension suffix.

Files Changed (8 files)

• api/core/rag/extractor/notion_extractor.py
• api/core/rag/extractor/word_extractor.py
• api/core/rag/extractor/markdown_extractor.py
• api/core/rag/extractor/extract_processor.py
• api/tests/unit_tests/core/rag/extractor/test_notion_extractor.py
• api/tests/unit_tests/core/rag/extractor/test_word_extractor.py
• api/tests/unit_tests/core/rag/extractor/test_markdown_extractor.py
• api/tests/unit_tests/core/rag/extractor/test_extract_processor.py