Summary

• expand MarkItDownOptions with detailed OCR controls (DPI, PSM, OEM, threads, deskew, color depth)
• preprocess images with optional scaling, grayscale, DPI metadata and gated deskew
• wire Tesseract engine to new options and thread limit
• enhance OcrBench with selective refresh, per-file logging and quality gates
• refresh markitdownnet OCR artifacts (eng, PSM=6, DPI=300)

Testing

• dotnet test
• dotnet run --project tools/OcrBench -- extract --input-dir dataset/validation --out-dir dataset/validation/_ocr --threads 1 --langs eng --psm 6 --refresh markitdownnet
• dotnet run --project tools/OcrBench -- compare --ocr-dir dataset/validation/_ocr --out-json artifacts/validation/OCR/bench-ocr.json --out-md artifacts/validation/OCR/summary-ocr.md (fails: GLOBAL Token-F1 0.7036 line_F1 0.3879)

https://chatgpt.com/codex/tasks/task_e_68a786d928ac8325bb19b5e3bce40bff