Extract documents from a URL into per-unit Markdown with metadata and OCR diagnostics. Useful for indexing PDFs, slides, sheets, scans, and mixed-format files.