Fix stale glossary-sensitive output adoption

[cursor]

Bug and impact

Untracked output_chunk*.md files were record-only by default even when their source chunk selected current glossary terms. A stale translation could be written into run_state.json as if it used the current glossary, causing future selective runs to skip it and ship inconsistent terminology. Alias-only glossary edits for already-selected terms were also not included in term hashes.

Root cause

run_state.plan() treated all existing outputs without records as safe to adopt unless --retranslate-untracked was passed, and glossary.term_hash() ignored aliases even though aliases are injected into chunk prompts.

Fix

• Re-translate untracked outputs whose source chunks select current glossary terms; only glossary-free untracked outputs remain record-only.
• Include sorted aliases in term hashes when aliases exist.
• Update SKILL/README docs and add regression coverage.

Validation

• python3 -m unittest tests.test_run_state tests.test_glossary -v
• python3 -m compileall scripts tests
• python3 -m unittest discover -s tests -p 'test_*.py' -v (209 tests, 3 skipped)
• CLI repro confirms stale glossary-sensitive output remains in translation_chunk_ids through record-only adoption, and alias-only changes trigger glossary_term_hash_changed.
• Standard Alice mock pipeline: 38 chunks merged; DOCX/PDF generated; EPUB failed only with the documented Ubuntu Calibre 7.6.0 bytes/str issue.