feat(markdown_parser): GFM HTML safe-list and footnotes in .md preview

[BunsDev]

Summary

• CodeView's .md Rendered preview now renders GFM raw HTML on a GitHub-style safe-list: <details>/<summary> (always-expanded with a ▾ glyph in v1), <kbd>, <sub>/<sup>, <del>/<ins>/<u>, raw <table>, <img>, plus inline phrasing tags. Tags outside the safe-list pass through as literal text; <script>/<style>/<iframe>/etc. strip.
• GFM footnotes ([^id] references + [^id]: defn definitions) resolve: references become numbered superscript-style hyperlinks (#fn-id); a horizontal rule + numbered list appears at the bottom with a back-link (#fnref-id); unused definitions are dropped.
• Mermaid (already wired) is verified via the included smoke fixture — no code change.

Spec lives at specs/castcodes-md-gfm-html/. All work is parser-only (crates/markdown_parser/); no editor / renderer / app changes. The bespoke <u> inline parser is removed in favor of routing through the same html5ever-backed helper. 161 unit tests pass.

Test plan

• Unit: `cargo test -p markdown_parser` — expect 161 passed, 0 failed.
• Downstream compile: `cargo check -p warp-app` — expect clean (no errors).
• Manual smoke test in CodeView (the deferred Phase 5.2 step):
  • Open `crates/markdown_parser/test-fixtures/gfm-smoketest.md` in the app.
  • Click the Rendered toggle in the markdown file header.
  • Verify each section renders as described in `specs/castcodes-md-gfm-html/PLAN-01-html-and-footnotes.md` §Task 5.2 (inline kbd/sub/sup/u/del/mark, block details with disclosure glyph, raw HTML table, mermaid SVG, footnotes section with back-link, `<script>` stripped, existing GFM still works).
• Lint: `cargo clippy -p markdown_parser --no-deps` — clean.

Deferred follow-ups (from final code review)

• Footnote rewriter only inspects each fragment in isolation; an inside-bold reference like `claim[^x]` may not match if the inline emphasis parser splits the bracket fragments differently than expected. Add coverage + fix if it reproduces.
• Footnote definition lines aren't gated on a preceding blank line, so an indented continuation can absorb the line following a mid-paragraph match.
• Footnote-reference fragments inside tables round-trip to `1` rather than `[^x]` via `inline_to_markdown` (TECH §7 already calls out the round-trip risk class).
• TECH §6 listed end-to-end `markdown_parser_tests` for several cases that ended up in `html_parser_tests` (helper-level) instead — add the end-to-end variants.

Notes

• 14 signed commits (every commit on this branch shows `Good "git" signature`).
• AI-attribution guard passes (`./script/check_ai_attribution`).
• No feature flag added — the change is a refinement of existing default-on markdown rendering. Previous behavior (raw HTML as literal text, `[^id]` as literal text) was a bug, not a feature to toggle.

[Copilot]

Pull request overview

This PR extends crates/markdown_parser so CodeView’s rendered .md preview supports (1) a GitHub-style safe-listed subset of raw HTML and (2) GFM footnotes, while keeping the public parser API stable and reusing the existing html5ever-based HTML pipeline.

Changes:

• Add a GFM HTML span lexer + wire block/inline HTML dispatch into the markdown parser.
• Add a footnote extract → rewrite → append pipeline around the existing markdown parse.
• Extend the HTML parser to render <details>/<summary>, <img>, raw <table>, and additional phrasing tags; add fixtures/spec docs and unit tests.

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
crates/markdown_parser/src/markdown_parser.rs	Runs footnote pre/post passes; adds block HTML dispatch loop; adds inline HTML token handling.
crates/markdown_parser/src/markdown_parser_tests.rs	Adds unit coverage for block/inline HTML behavior and stripping.
crates/markdown_parser/src/html_parser.rs	Exposes pub(crate) HTML fragment helpers; adds <details>, <img>, raw <table>, and more phrasing-tag styling.
crates/markdown_parser/src/html_parser_tests.rs	Adds regression coverage for fragment parsing + new HTML constructs.
crates/markdown_parser/src/gfm_html.rs	Introduces safe/strip tag lists and a span lexer for HTML in markdown.
crates/markdown_parser/src/gfm_html_tests.rs	Adds lexer unit tests (nesting, attributes, comments, malformed tags).
crates/markdown_parser/src/footnotes.rs	Implements footnote definition extraction, reference rewriting, and section appending.
crates/markdown_parser/src/footnotes_tests.rs	Adds unit coverage for extraction and end-to-end footnote rendering behavior.
crates/markdown_parser/src/lib.rs	Registers new private modules (footnotes, gfm_html).
crates/markdown_parser/test-fixtures/gfm-smoketest.md	Adds a manual smoke fixture for in-app visual verification.
DESIGN-CHANGES.md	Documents the user-visible preview behavior changes.
specs/castcodes-md-gfm-html/TECH.md	Adds the technical design/spec for HTML safe-list + footnotes implementation.
specs/castcodes-md-gfm-html/PRODUCT.md	Adds product requirements and expected rendering behavior.
specs/castcodes-md-gfm-html/PLAN-01-html-and-footnotes.md	Adds an implementation plan and verification checklist.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.