another fix: push checkout/installation into if block and add vault setup for custom E2E reporter

[Dave Shoup (shouples)]

Good catch by Copilot #3340 (review)

[James Robinson (jlrobins)]

yeah, don't do half work unless going to also do the other half, eh?

[Copilot]

Pull request overview

Adjusts the Semaphore Playwright E2E pipeline’s after_pipeline reporting job so it only performs repository checkout / dependency installation / Vault setup when the workflow is schedule-triggered, enabling the custom Playwright reporter to be available and authenticated.

Changes:

• Wrap checkout, cache restore, npm ci, and Vault setup inside the scheduled-run if block in after_pipeline.
• Add Vault setup before running the custom Kafka Playwright results reporter during scheduled runs.
• Minor YAML formatting change to the blob-report artifact push command.

Comments suppressed due to low confidence (1)

.semaphore/playwright-e2e.yml:69

• blob-report/*.zip can expand to multiple files, but $(basename blob-report/*.zip) and artifact push ... --destination ... expect a single destination path. This will fail or upload to an unintended destination when multiple blob report zips exist (the make target docs indicate multiple zips are expected). Consider looping over each zip (computing basename per file) or using an artifact destination directory that preserves filenames.

            - artifact push workflow blob-report/*.zip --destination blob-report/$(basename blob-report/*.zip)
            - make remove-test-env

[Copilot]

vault-setup is sourced unconditionally inside the scheduled-run block. Earlier in this pipeline the Vault setup is guarded by checking SEMAPHORE_ORGANIZATION_URL to skip Vault on Semaphore CI; without the same guard here, scheduled runs on hosted Semaphore could fail during after_pipeline. Consider reusing the same conditional skip logic before sourcing vault-setup.

Suggested change

	. vault-setup
	if [[ -z "${SEMAPHORE_ORGANIZATION_URL:-}" ]]; then
	. vault-setup
	fi

[Dave Shoup (shouples)]

No need for this since scheduled runs aren't set up for the public Semaphore instance