fix: wire workflow_call inputs, act fallback for change detection, and correct clang-tidy/format-all failure behavior

[Copilot]

Several behavioral regressions introduced by the workflow consolidation: workflow_call inputs were not forwarded to workflow-setup, causing reused workflows to resolve the wrong ref/repo/base SHA; change detection ran unconditionally in act environments causing unreliable results; and clang-tidy-check/format-all no longer failed their jobs when issues were found.

workflow-setup — act environment handling

• Gate Run change detection on is_act != 'true'
• Add Set has_changes output step: returns true under act (so check jobs run locally), otherwise passes through detection result
• Update has_changes output to reference new step

Workflow call input forwarding

Pass ref, repo, pr-base-sha, checkout-path from workflow_call inputs into workflow-setup for:

• python-check.yaml, markdown-check.yaml, jsonnet-format-check.yaml
• cmake-format-check.yaml, header-guards-check.yaml, actionlint-check.yaml

Without this, reused workflows fell back to the caller event context instead of the explicit inputs, breaking relevance detection and checkout targeting.

format-all.yaml — restore failure behavior

Add Fail on formatter failures step (runs after comment/reaction steps) so combine-results exits non-zero when any sub-workflow fails/cancels/is skipped. Restores parity with the pre-consolidation inline script.

clang-tidy-check.yaml — actually fail on diagnostics

Add exit 1 when clang-tidy issues are detected. The Upload artifacts step already uses if: always() so artifacts are still collected on failure.

---

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

[Copilot]

Pull request overview

This PR fixes behavioral regressions introduced by the workflow consolidation by ensuring reusable (workflow_call) workflows correctly forward ref/repo/base SHA inputs into workflow-setup, improving act behavior for change detection, and restoring “fail the job when issues are found” semantics for formatter/tidy workflows.

Changes:

• Forward workflow_call inputs (ref, repo, pr-base-sha, checkout-path) into workflow-setup across several check workflows so ref/base SHA resolution works correctly when reused.
• Update workflow-setup to skip change detection under act and provide a has_changes output that allows check jobs to run locally.
• Restore failure behavior in format-all aggregation and make clang-tidy-check fail the job when diagnostics are detected.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
.github/workflows/python-check.yaml	Forward workflow_call inputs into workflow-setup for correct ref/repo/base SHA resolution.
.github/workflows/markdown-check.yaml	Same input forwarding to ensure reusable workflow resolves correct checkout/ref context.
.github/workflows/jsonnet-format-check.yaml	Same input forwarding for correct checkout targeting and relevance detection inputs.
.github/workflows/header-guards-check.yaml	Same input forwarding for reusable invocations.
.github/workflows/cmake-format-check.yaml	Same input forwarding for reusable invocations.
.github/workflows/actionlint-check.yaml	Same input forwarding for reusable invocations (with include-globs-based relevance).
.github/actions/workflow-setup/action.yaml	Skip change detection under act and standardize has_changes via a dedicated step.
.github/workflows/format-all.yaml	Fail combine-results when any formatter workflow fails/cancels/is skipped (restores prior behavior).
.github/workflows/clang-tidy-check.yaml	Exit non-zero when clang-tidy diagnostics are detected so the job actually fails.

---

You can also share your feedback on Copilot code review. Take the survey.

[Copilot]

Set has_changes output can write an empty has_changes= when change detection is not run (e.g., file-type and include-globs are both empty so steps.detect is skipped). For consistency (and to keep the output strictly boolean-like), consider explicitly emitting has_changes=false in that case (while keeping the act override behavior).

Suggested change

	else
	echo "has_changes=${{ steps.detect.outputs.has_changes }}" >> "$GITHUB_OUTPUT"
	elif [ "${{ steps.detect.outputs.has_changes }}" != "" ]; then
	echo "has_changes=${{ steps.detect.outputs.has_changes }}" >> "$GITHUB_OUTPUT"
	else
	echo "has_changes=false" >> "$GITHUB_OUTPUT"

[Copilot]

This branch emits a workflow command as ::warning::... but then exits non-zero, which fails the job. Consider switching this to ::error::... (or keeping warning but not failing) so the log annotation severity matches the job outcome.

Suggested change

	echo "::warning::Clang-tidy found issues in the code"
	echo "::error::Clang-tidy found issues in the code"