feat: extract TextContent from text, JSON, and XML files

[davebartlett442]

Summary

• Add a native text processor for any text/* MIME, plus JSON (application/json, +json) and XML (application/xml, +xml)
• Decode source bytes as UTF-8 (BOM stripped, lossy fallback) and emit TextContent plus a single searchable page
• Leave LibreOffice conversion first in the dispatch order so text/html, text/spreadsheet, and application/xhtml+xml stay on the existing PDF path

Test plan

• Unit tests in packages/docbox-processing/tests/test_process_text.rs cover plain text, markdown, csv, JSON, XML, BOM stripping, empty files, and unrelated MIME types
• CI docbox-processing tests on this PR
• Upload a .txt, .md, .json, and .xml file and confirm generated TextContent plus search indexing
• Confirm HTML/office files still convert through LibreOffice as before

Made with Cursor

[jacobtread]

Have opened another PR expanding on this with some changes. #15 have moved these changes over to there