Fix ref input, eyes reaction handling, and docs consistency in fix workflows

[Copilot]

Review feedback on the format-all consolidation PR identified several issues: fix workflows accepting commit SHAs as ref (which breaks git push origin HEAD:<ref>), docs examples using head.sha instead of head.ref, a fragile eyes-reaction removal that could pass undefined as reaction_id, and a misleading patch artifact message.

Fix workflow ref input

Updated workflow_call ref description in all 7 fix workflows to make the branch-name requirement explicit:

ref:
  description: "The branch name to checkout and push fixes to (must be a branch, not a commit SHA)"

Documentation

• All fix workflow usage examples updated from head.sha to head.ref
• All All Inputs ref entries updated to reflect the branch-name requirement

Remove eyes reaction in format-all.yaml

Replaced the inline chained call (which passed undefined to deleteForIssueComment when no eyes reaction existed) with an explicit fetch-then-conditionally-delete pattern:

const { data: reactions } = await github.rest.reactions.listForIssueComment({ ... });
const eyesReaction = reactions.find(
  (r) => r.content === 'eyes' && r.user && r.user.type === 'Bot'
);
if (eyesReaction && eyesReaction.id) {
  await github.rest.reactions.deleteForIssueComment({ ..., reaction_id: eyesReaction.id }).catch(() => {});
}

Patch artifact message

Clarified that the patch file (fix.patch) lives inside the fix-patch artifact, not that the artifact is named after the patch file.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[Copilot]

Pull request overview

This PR tightens up reusable “fix” workflow documentation/inputs so callers pass a branch ref (not a SHA), and hardens format-all.yaml’s reaction cleanup and patch-artifact messaging.

Changes:

• Updated workflow_call ref input descriptions across fix workflows to require a branch name (not a commit SHA).
• Updated reusable workflow documentation examples and input docs to use pull_request.head.ref and describe the branch-name requirement.
• Refactored format-all.yaml to safely find-and-delete the “eyes” reaction and clarified the patch artifact message.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
.github/workflows/yaml-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/python-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/markdown-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/jsonnet-format-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/header-guards-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/cmake-format-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/clang-format-fix.yaml	Clarifies workflow_call.inputs.ref as a branch name.
.github/workflows/format-all.yaml	Makes eyes-reaction deletion conditional and improves patch messaging.
.github/REUSABLE_WORKFLOWS.md	Updates examples/inputs to use head.ref and document branch-name ref.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 1 comment.

[Copilot]

The listForIssueComment call is not guarded. If the API call fails (permissions/transient errors), this step will throw and can fail combine-results, preventing the combined PR comment from being posted. Since reaction cleanup is best-effort, wrap the list+delete logic in a try/catch (or mark the step continue-on-error: true) so the workflow can continue even if reactions can’t be queried.