ci(e2e): restart job service after DB setup completes

The knative-job-service container boots in parallel with the kind
cluster and uses async-retry (factor: 2) to wait for app_jobs.jobs to
exist. When an early retry lands while the schema is still being seeded
by the constructive-db Job, the next retry is scheduled ~70s later -
longer than the e2e test's 30s waitForJobComplete window, causing
flakes (most recently E2E simple-email on PR #38).

Force-replace the pod after the DB setup Job completes so the new pod's
first prereqs check succeeds immediately. Adds ~3-8s per matrix entry
and removes the race.

Author: Anmol1696

.github/workflows/test-k8s-deployment.yaml

@@ -139,6 +139,11 @@ jobs:
           kubectl wait --for=condition=complete job/constructive-db \
             -n constructive-functions --timeout=180s
 
+      - name: Restart job service after DB ready
+        run: |
+          kubectl rollout restart deploy/knative-job-service -n constructive-functions
+          kubectl rollout status deploy/knative-job-service -n constructive-functions --timeout=120s
+
       - name: Port-forward postgres and run e2e tests
         env:
           PGHOST: localhost