feat(ingest): bulk_ingest.py + run_audit.py DOCX/HTML/detector fixes

bulk_ingest.py — new script for bulk multi-jurisdiction DB ingestion:
  - Auto-maps folder names (DINU-ODIA→dinuba, EXET_ODIA→exeter, etc.)
  - Supports PDF, DOCX, DOC, TXT, JSON, XML, HTML/HTM
  - Skips audio/image/binary files automatically
  - Deduplicates by document_id (safe to re-run)
  - --dry-run flag to preview finding counts without writing to DB
  - --folder flag to process a single jurisdiction subfolder
  - Persists Document + Analysis + Anomaly rows to oraculus_audit.db

run_audit.py fixes:
  - DOCX/DOC support via python-docx
  - HTML/HTM support via BeautifulSoup
  - Replace stale _SINGLE_DOC_DETECTORS with analyze_document()
  - Add --jurisdiction flag
  - Restore logger definition

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: MARS