Match link labels with Unicode case folding

[bobzhang]

Split out of #152.

Bug: link labels were only lowercased for ASCII. The spec matches labels after Unicode case folding.

Repro:

[ΑΓΩ]: /φου

[αγω]

gave <p>[αγω]</p> instead of a link. Same for [ẞ] vs [SS]: /url.

Fix: link_label case-folds the normalized label with @casefold.casefold (full Unicode case folding; moonbit-community/casefold is already a module dependency, now imported by cmark_base).

Spec examples fixed: 206, 540.

Tests: src/cmark_html/link_label_casefold_test.mbt.

Depends on: nothing, independent of the other PRs.

[bobzhang]

Automated review: Codex CLI (codex exec, model reasoning effort high, read-only sandbox), 1 round, on git diff against the base branch.

Verdict: signed off, no changes requested.

I sign off. No concrete correctness or regression issues found in origin/main...HEAD.

The shared parser applies full Unicode folding consistently to definitions, reference links, and images, matching CommonMark 0.31.2. Folding preserves source positions and does not affect the existing length counter. Tests cover Greek and expanding folds (ẞ, ß).

Codex could not run moon test in its read-only sandbox. The tests were run separately on native, js and wasm-gc and pass.