fix(lint): fail lint:fix when unfixable errors remain

[danyalahmed1995]

Summary

Fixes npm run lint:fix so it exits with a failure when unfixable lint errors remain after the fixer runs.

Previously, lint:fix could successfully apply available fixes and exit 0, even when a follow-up npm run lint would still fail on the same target files. This made the command report success for cases it could not fully fix.

The fix reuses the normal lint pass after running the existing fixers and exits non-zero when lint errors remain.

Fixes #29615.

Validation

• npm run lint:fix -- http/headers/X-Frame-Options.json
• npm run lint -- http/headers/X-Frame-Options.json
• Temporarily reproduced the missing spec_url case from Fixer does not report "Marked as standard_track, but missing required spec_url" #29615 and confirmed lint:fix now exits with failure when the remaining lint error is unfixable.
• npm run format
• npm run lint
• Unit tests passed with the PowerShell-compatible equivalent after the npm script hit Windows NODE_ENV shell syntax.

[github-actions]

Tip: Review these changes grouped by change (recommended for most PRs), or grouped by feature (for large PRs).

[danyalahmed1995]

System file check failed because I changed the lint/fix.js 🫠

[caugner]

This is an interesting workaround that I hadn't thought about, but unfortunately it would significantly increase the time it takes to run fix:lint.

I will bring this up in tomorrow's BCD meeting.

[danyalahmed1995]

Yes you are right, I kept this intentionally small to match the current behavior gap, but I can see how re-running the full lint pass after lint:fix could be too expensive for regular use.

I will look for a better solution and get back to you.

[danyalahmed1995]

There are couple of options we can take a look at

1. Keep this as-is, but accept the extra lint pass on the same target files.
2. Add a strict mode such as --fail-on-unfixed, so normal lint:fix stays fast while pre-commit/CI can fail when unfixable errors remain.
3. Move the follow-up lint call only into the pre-commit path, since that is where the issue was observed.
4. Run only a narrower post-fix validation for unfixable checks like the standard_track / spec_url case.
5. Do a larger linter/fixer unification so fixable and unfixable diagnostics come from one shared pass.

Let me know what you think.