Fix flaky TestDestroyPodInflight test

This fix deletes Route before Configuration to prevent reconciliation
errors during teardown (fixes "revision not found" and "failed to fetch
loadbalancer" errors in CI logs)

The test now correctly validates that in-flight requests complete
successfully during graceful shutdown.

Author: Fedosin

test/e2e/destroypod_test.go

@@ -144,8 +144,16 @@ func TestDestroyPodInflight(t *testing.T) {
 		// Give the request a bit of time to be established and reach the pod.
 		time.Sleep(timeoutRequestDuration / 2)
 
-		t.Log("Destroying the configuration (also destroys the pods)")
-		return clients.ServingClient.Configs.Delete(egCtx, names.Config, metav1.DeleteOptions{})
+		t.Log("Destroying the route and configuration (also destroys the pods)")
+		// Delete the Route first to prevent reconciliation errors during cleanup
+		if err := clients.ServingClient.Routes.Delete(egCtx, names.Route, metav1.DeleteOptions{}); err != nil {
+			return fmt.Errorf("failed to delete route: %w", err)
+		}
+		// Then delete the Configuration
+		if err := clients.ServingClient.Configs.Delete(egCtx, names.Config, metav1.DeleteOptions{}); err != nil {
+			return fmt.Errorf("failed to delete config: %w", err)
+		}
+		return nil
 	})
 
 	if err := g.Wait(); err != nil {