fix after_pipeline for E2E reporting

[Dave Shoup (shouples)]

Follow-up based on the errors we're seeing in CI For the scheduled E2E test runs:

[Mar 23 07:56:31.361] Successfully pulled artifact for current workflow.
[Mar 23 07:56:31.361] * Remote source: 'artifacts/workflows/33e65626-f701-4c4e-a707-6b8b066c142c/blob-report'.
[Mar 23 07:56:31.361] * Local destination: 'blob-report'.
[Mar 23 07:56:31.361] Pulled 8 files. Total of 172.6 MB
npm warn exec The following package was not found and will be installed: playwright@1.58.2
Error: Cannot find module './tests/e2e/reporters/kafka-results-reporter.ts'

• need checkout so tests/e2e/reporters/kafka-results-reporter.ts is available
• make ci-bin-sem-cache-restore isn't really necessary, but should help speed up npm ci
• npm ci added since the custom reporter needs to use a few packages

[Copilot]

Pull request overview

Fixes scheduled Semaphore Playwright E2E after_pipeline reporting by ensuring the custom Kafka reporter source and its Node dependencies are available when merging blob reports.

Changes:

• Add checkout in after_pipeline so tests/e2e/reporters/kafka-results-reporter.ts exists when referenced.
• Restore Semaphore caches and run npm ci in after_pipeline so the reporter’s runtime dependencies are installed.

[Copilot]

kafka-results-reporter.ts reads credentials via vault kv get at module load time, but this after_pipeline job doesn't run the same Vault setup used by the E2E blocks (the conditional . vault-setup/login). Once the reporter file is available, npx playwright merge-reports is likely to fail here due to missing Vault auth/token. Add the same Vault initialization step in this job (ideally only for the scheduled path).

Suggested change

	- if [[ "$SEMAPHORE_WORKFLOW_TRIGGERED_BY_SCHEDULE" == "true" ]]; then
	- if [[ "$SEMAPHORE_WORKFLOW_TRIGGERED_BY_SCHEDULE" == "true" ]]; then
	. vault-setup;

[Copilot]

checkout, cache restore, and npm ci now run unconditionally in after_pipeline, but the blob report merge (and Kafka publishing via the custom reporter) only runs for scheduled workflows. Consider moving these expensive steps inside the SEMAPHORE_WORKFLOW_TRIGGERED_BY_SCHEDULE conditional so non-scheduled runs don’t pay the extra time/network cost.