Recover named references and truncated HTML text

[rogerchappel]

Summary

• decode semicolon-terminated HTML Latin-1 named references while leaving semicolonless and unknown names unchanged
• omit comments that run to end-of-input and recognize only syntactically tag-shaped markup, preserving ordinary </> comparisons
• add focused unit coverage, an executable recovery fixture, and aligned API/safety documentation

Commits

• 0fa1dab test: define deterministic HTML recovery
• 24fb68e fix: recover common HTML text edge cases
• 940aaa8 docs: exercise supported HTML recovery

Verification

• npm run release:check (36 tests, syntax, build, CLI smoke, package smoke, release contract)
• bash scripts/validate.sh
• git diff --check origin/main...HEAD

All commits are authored and committed by Roger Chappel miscanalysis@gmail.com.

[rogerchappel]

Automated merge note

Triage class: auto-merge

Summary: deterministic local HTML-to-text recovery for named entities, truncated comments, and text comparisons (8 files, +46/-7).

Checks run: CI test completed SUCCESS; required review/branch protection re-read.

Rebased/CI-repaired: no.

Verified head SHA: 940aaa8b74aa9bfd91e1fe22a6599f2205d0f9ed.