Image OCR ingestion, import/share integration, license, and transitive-dependency policy

[etherlink-intern]

Motivation

• Make images first-class inputs by adding on-device OCR so apps can import photos and scans and convert them to Markdown on Apple platforms.
• Provide clear integration guidance for common import surfaces (document picker, PhotosPicker, and Share Sheet) so apps can wire a single conversion core into multiple UI flows.
• Make dependency/acknowledgement policy rigorous by requiring transitive SwiftPM dependencies to be recorded for future FOSS backends (e.g., CoreXLSX → XMLCoder/ZIPFoundation).

Description

• Added image document formats to DocumentFormat (png, jpeg, heic, tiff, gif) and an imageFormats helper, extended content-type and filename inference for images.
• Implemented ImageOCRConverter that uses Vision/CoreGraphics/ImageIO under conditional compilation to run OCR, normalize recognized lines, and return MarkdownDocument, and registered it in MarkItDown.defaultConverters.
• Added comprehensive docs and examples: Docs/ImportButtonIntegration.md (file importer, PhotosPicker, Share Extension, importer helpers) and updated Docs/NativeConverterBackends.md and README.md to document planned backends, platform availability, and the new transitive-license acknowledgement policy (including adding XMLCoder to the acknowledgement table).
• Added repository LICENSE (MIT), added tests for format inference, platform fallback behavior, and conditional Vision OCR (rendered OCR test images when Vision is available), and updated the test suite to exercise the new image paths.

Testing

• Ran swift test, which completed successfully (unit tests passed, Vision-based tests run conditionally).
• Ran the CLI smoke tests via Scripts/smoke-test.sh, which completed successfully.
• Ran repository checks via git diff --check (no issues found).

---

Codex Task