pdf_oxide 0.2.2

Production-grade PDF parsing: spec-compliant text extraction, intelligent reading order, OCR support. 47.9× faster than PyMuPDF4LLM.