Decode legacy entities without semicolons

[rogerchappel]

Summary

• add table-driven regression coverage for legacy semicolonless named references and ambiguous boundaries
• decode the explicit legacy text subset without consuming adjacent alphanumeric text
• add an inspect-path fixture and document supported boundary behavior

Commits

• 5e6ef53 test: cover legacy semicolonless entities
• c4f4c21 fix: decode legacy entities without semicolons
• b8713da docs: define legacy entity boundaries

Verification

• npm ci --ignore-scripts --no-audit --no-fund
• npm run release:check (47/47 tests, syntax checks, build, CLI smoke, package smoke, release contract)
• git diff --check origin/main...HEAD

The pre-fix suite failed the new entity, fixture-inspection, and CLI-inspection regressions as expected.

[rogerchappel]

Automated merge note

Triage class: auto-merge

Summary: Adds safe legacy semicolonless HTML entity decoding at non-alphanumeric boundaries, with fixture, unit, CLI, and documentation coverage (8 files, +50/-9).

Checks run: GitHub CI test — COMPLETED/SUCCESS; final live diff, mergeability, branch protection, and Roger Chappel author/committer identities re-verified.

Rebased/CI-repaired: No.

Verified head SHA: b8713da5d5f426ec4beb98e9453e1aa2957faec0.