Handle raw HTML embedded in Markdown

[max-heller]

Currently, raw HTML tags in the Markdown source are usually ignored or passed through to Pandoc, which ignores them. This doesn't work well for a few reasons:

• ID attributes are ignored, resulting in broken links
• Paths (href, src, etc.) aren't normalized/rebased
• HTML elements like <dl>, <dt>, and <dd> are ignored (see Render book as PDF in publish.yml workflow google/comprehensive-rust#1572 (comment))

Options:

• Force Pandoc to parse raw HTML
  • This doesn't handle raw HTML inlines e.g. <a href="foo"></a>
  • Filters run after the reader is finished, so this won't work for HTML containing links (they need to be corrected with --file-scope)

Considerations:

• Parsing should happen before mdbook-pandoc does its preprocessing pass. Accomplishing this without running Pandoc once per HTML section in each source file seems difficult
• Parsing should happen before Pandoc is run with --file-scope so links in raw HTML are adjusted correctly

See also:

• feature request: option to have html parsed inside markdown jgm/pandoc#2716 (comment)