Fix potential panic in getMachineConfigOperatorPod when no pods are returned

Summary

In test/extended-priv/util.go, the helper function getMachineConfigOperatorPod indexes pods[0] before checking whether exutil.GetAllPodsWithLabel succeeded or returned any pods. This can cause a runtime panic if the label query errors or returns an empty list.

Related

• PR: https://github.com/openshift/machine-config-operator/pull/5875
• Review comment: https://github.com/openshift/machine-config-operator/pull/5875#discussion_r3128857555
• Jira: https://redhat.atlassian.net/browse/MCO-2243
• Requested by: @ptalgulk01

Proposed Fix

func getMachineConfigOperatorPod(oc *exutil.CLI) (string, error) {
	pods, err := exutil.GetAllPodsWithLabel(oc.AsAdmin(), MachineConfigNamespace, "k8s-app=machine-config-operator")
	if err != nil {
		return "", err
	}
	if len(pods) == 0 {
		return "", fmt.Errorf("no machine-config-operator pod found in namespace %s", MachineConfigNamespace)
	}
	logger.Infof("machine-config-operator pod name is %s", pods[0])
	return pods[0], nil
}

Start in test/extended-priv/util.go at getMachineConfigOperatorPod and inspect how its callers handle the returned error. Verify the helper handles both GetAllPodsWithLabel errors and an empty pod list without indexing pods[0], while still returning the pod name when one is found.