Add markdown file and directory support

[brianmatzelle]

Summary

• Open single .md files as one-chapter books (e.g. bookokrat README.md)
• Open directories containing .md files as multi-chapter books with TOC (e.g. bookokrat wiki/)
• Markdown directories get proper chapter ordering: Home.md/README.md/index.md sorted first, then alphabetical
• Uses pulldown-cmark for GFM-compatible conversion (tables, strikethrough, task lists)
• Markdown files and directories appear in the library browser alongside EPUBs and PDFs

How it works

Follows the same pattern as the existing HTML file support — markdown content is converted to HTML via pulldown-cmark, then wrapped in a temporary EPUB that the existing rendering pipeline handles. For directories, each .md file becomes a chapter with a proper toc.ncx for navigation.

No changes to the rendering pipeline, text reader, or PDF infrastructure.

Changes

• Cargo.toml — add pulldown-cmark dependency
• src/book_manager.rs — BookFormat::Markdown variant, format detection for .md files and markdown directories, markdown_to_html(), create_fake_epub_from_markdown() (single file), create_epub_from_markdown_dir() (multi-chapter), 6 new unit tests
• src/main_app.rs — route BookFormat::Markdown to the EPUB loader
• src/main.rs — update help text

Test plan

• All 385 existing tests pass (zero regressions)
• 6 new unit tests covering: format detection for .md files, format detection for markdown directories, single file loading with HTML conversion verification, multi-chapter directory loading with chapter count and ordering verification, library discovery of .md files, library discovery of markdown directories
• Manual testing: bookokrat path/to/file.md opens and renders correctly
• Manual testing: bookokrat path/to/wiki/ opens as multi-chapter book with TOC sidebar

🤖 Generated with Claude Code

[brianmatzelle]

Hi @bugzmanov !

I generated this feature with Claude Code, so I could read my project's Markdown wikis with bookokrat.

It works great for me locally. Completely understand if you'd rather close this PR though to keep bookokrat centered around .epubs

[bugzmanov]

@brianmatzelle i like this feature! (i'm reading more markdown files than books these days 😩)

But design wise i don't like new dependency (pulldown-cmark). And the proposed design makes the processing pipeline a bit weird: Markdown → HTML → EPUB → HTML → Markdown AST → render. too many extra steps that looks unnecessary.

What do you think about implementing a markdown parser? that transforts text to custom markdown AST (src/markdown.rs) ?
Potentially i would reuse that feature to support markdown-based comments in the future :)

(+ branch has conflicts with main)

[brianmatzelle]

I agree that this implementation is a bit weird since it was vibe coded.

Sure, I'll look into it with more detail and see what I can do. Less vibe code and more whiteboard, ya know?