fix(ingestor): ParseError(ValueError) escapes as HTTP 500 for encrypted/oversized PDFs #282

Summary

ParseError(ValueError) raised by PyMuPDFParser.parse() for encrypted or oversized PDFs is caught at ingestor.py:214 by except (AssertionError, ValueError, TypeError): raise, which re-raises it bare. FastAPI has no handler for raw ValueError, so callers receive HTTP 500.

Root cause

Fix #242/#246 added class ParseError(ValueError) in parser.py. But PipelineIngestor at line 214 has except (AssertionError, ValueError, TypeError): raise which catches ParseError (since it IS a ValueError) and re-raises before it can be mapped to a user-friendly error.

Fix

backend/app/adapters/ingestor.py -- catch ParseError before the generic ValueError catch and map it to a 422:

• from ingestion.parser import ParseError
• except ParseError as exc: raise validation_error(str(exc)) from exc
• (then the existing ValueError block re-raises)

Impact

Users uploading encrypted PDFs or files >100 MB receive HTTP 500 instead of 422.