Add AKF support for EPUB format

[HMAKT99]

Summary

AKF currently embeds metadata into 20+ formats (DOCX, PDF, images, HTML, etc.) but doesn't yet support EPUB — the standard ebook format.

What needs to happen

1. Add an embed_epub() function in python/akf/embed.py
2. EPUB files are ZIP archives containing XHTML — metadata can go in the OPF metadata section or as a custom <meta> tag
3. Add corresponding extract_epub() in python/akf/extract.py
4. Add tests in python/tests/test_embed.py

Helpful context

• Look at embed_docx() for a similar ZIP-based format approach
• EPUB spec: https://www.w3.org/publishing/epub3/epub-packages.html
• The metadata should survive round-trips (embed → extract → verify identical)

Acceptance criteria

• akf embed book.epub works
• akf extract book.epub recovers the metadata
• Tests pass