Is there a way to fail early if a task in a k8s pipeline references a queue that doesn't actually exist? We've had this happen by accident (typo), the pipeline just kept running indefinitely.