What and why?

The Cloud Run e2e tests previously only verified resource shape (sidecar injection, env vars, volume mounts) using the stock hello image, which emits no telemetry. This PR extends them to verify that spans and logs from the instrumented app actually reach the Datadog e2e org.

Note, we may want to migrate to dd-sts at some point to avoid having api/app keys stored in github

How?

• Swaps the deploy image to gcr.io/datadog-serverless-gcp-dev/run-nodejs-sidecar (a real Node.js app with tracing), controlled via the new GCP_CLOUD_RUN_APP_IMAGE_E2E repo variable
• Switches to --allow-unauthenticated so the test can hit the service URL without credentials
• Adds --tracing to the instrument invocation to ensure DD_TRACE_ENABLED=true
• Adds a new telemetry flows test case (between instrument and uninstrument) that:
  1. Resolves the service URL via gcloud run services describe
  2. Hits the endpoint with fetch to generate a trace
  3. Polls the DD v2 Spans and Logs APIs in parallel (20 × 15s, 5 min max) until both return results filtered by service name
• Adds e2e/helpers/cloud-run-telemetry-checker.ts (mirrors the pattern from junit-upload-checker.ts) with explicit DATADOG_API_KEY/DATADOG_APP_KEY auth -- the e2e org keys already flow in from DATADOG_API_KEY_E2E/DATADOG_APP_KEY_E2E in CI
• Threads GCP_CLOUD_RUN_APP_IMAGE_E2E through CI env and adds the helper file to the cloud-run paths filter

Review checklist

• Feature or bugfix MUST have appropriate tests (unit, integration)