feat: convert OFD documents to Markdown #208

Problem and use case

anydoc does not recognize .ofd documents. OFD is a fixed-layout document format used for forms, invoices, and official records. Extracting only raw text loses headings, emphasis, lists, tables, and links that matter in Markdown.

Expected behavior

• Detect OFD from its package descriptor, including encrypted OFD packages, and recognize .ofd by extension.
• Convert page and template text in document order through the existing Document model.
• Preserve styles where the OFD provides reliable evidence: outline headings, larger-font headings, bold/italic runs, printed lists, complete bordered table grids, and URI links.
• Return NeedsOcr for image-only pages and Encrypted for encrypted packages instead of silently returning incomplete content.
• Expose the format through the Rust, Node.js, Python, and WebAssembly APIs without adding ofdrw as a runtime dependency.
• Cover format detection, Markdown output, multi-page ordering, Unicode escapes, tables, links, lists, and error paths with tests.

Conversion limits

OFD describes fixed-page drawing instructions rather than semantic document blocks. Heading size and bordered-table detection are necessarily conservative. Image content, vector drawings, non-link annotations, and OCR are outside the initial conversion scope; documents without enough layout evidence should retain text as paragraphs rather than inventing structure.