Allow three long term flaky k8s tests to retry

We attempted fixing in 4.20 but couldn't quite make it. All are believed
related to the kubelet metrics problem. Restoring retries for 4.20
release but the bugs will remain open.

Author: dgoodwin

pkg/test/ginkgo/cmd_runsuite.go

@@ -156,6 +156,21 @@ func shouldRetryTest(ctx context.Context, test *testCase, permittedRetryImageTag
 
 	tlog := logrus.WithField("test", test.name)
 
+	// Test retries were disabled for some suites when they moved to OTE. This exposed small numbers of tests that
+	// were actually flaky and nobody knew. We attempted to fix these, a few did not make it in time. Restore
+	// retries for specific test names so the overall suite can continue to not retry.
+	retryTestNames := []string{
+		"[sig-instrumentation] Metrics should grab all metrics from kubelet /metrics/resource endpoint [Suite:openshift/conformance/parallel] [Suite:k8s]", // https://issues.redhat.com/browse/OCPBUGS-57477
+		"[sig-network] Services should be rejected for evicted pods (no endpoints exist) [Suite:openshift/conformance/parallel] [Suite:k8s]",               // https://issues.redhat.com/browse/OCPBUGS-57665
+		"[sig-node] Pods Extended Pod Container lifecycle evicted pods should be terminal [Suite:openshift/conformance/parallel] [Suite:k8s]",              // https://issues.redhat.com/browse/OCPBUGS-57658
+	}
+	for _, rtn := range retryTestNames {
+		if test.name == rtn {
+			tlog.Debug("test has an exception allowing retry")
+			return true
+		}
+	}
+
 	// Get extension info to check if it's from a permitted image
 	info, err := test.binary.Info(ctx)
 	if err != nil {