[wanjinhao1]

What

Adds two lint rules to packages/lint/src/rules/composition.ts that surface the parser's invalid-duration and invalid-track-index diagnostics as lint errors (invalid_data_duration, invalid_track_index), with the element and the bad attribute value in the message.

Why

The parser already marks values like data-duration="2s" or "-3" as invalid-duration (compositionContract.ts ~L268), but the linter only turned deprecated-layer/deprecated-end/conflicting-end into findings, so these values linted clean and the clip silently got a different duration than the author wrote.

Fixes #7272

Related work

Refs #7272 (mentor @jrusso1020). Separate from #7291 (different rule, different issue).

How

New rule block iterating tags, reading each tag's timing via readTagTiming, and mapping invalid-duration / invalid-track-index diagnostics to error findings following the existing deprecated_data_layer/deprecated_data_end pattern (elementId, fixHint, snippet). No parser changes.

Test plan

• Unit tests added/updated: three new tests in composition.test.ts — invalid "2s" and "-3" durations report two invalid_data_duration errors; a valid 3.5 duration reports none; a 1.5 track index reports invalid_track_index. npx vitest run packages/lint/src/rules/composition.test.ts -t "invalid timing" → 3 passed.
• Manual testing performed — not run: the 42 invalid_composition_variables_declaration failures and the lefthook typecheck failure (missing generated runtime-inline, bun unavailable locally) reproduce on a clean checkout of origin/main and are unrelated to this change.
• Documentation updated — no docs changes needed; codes follow the existing snake_case finding scheme.
• Comments follow CONTRIBUTING.md "Comments".

This PR was prepared with AI assistance and reviewed by the contributor.

[jrusso1020]

Thanks for contributing this fix. I reviewed the competing submissions and selected #7286, which includes the parser-based duration checks, element/value/severity regression cases, and documentation for both new lint codes. Closing this as a duplicate so the issue has one implementation to take through CI and merge: #7286. This is consolidation of overlapping work, not a rejection of your contribution.