Currently our watchers break entirely if any object fails to deserialize. Ideally this should be confined to only affect the object in question.

The prerequisites have been done by fixing kube-rs and bringing the necessary changes into operator-rs. A sample implementation has also been done for NiFi, we now need to roll this out across the board.

• Watcher breaks if any object cannot be deserialized kube-rs/kube#774
• Allow other NifiClusters to reconcile even if one is invalid nifi-operator#662
• fix: don't stop the reconciliation if one cluster is invalid airflow-operator#520
• fix: don't stop reconciliation if one cluster is invalid druid-operator#638
• fix: don't stop the reconciliation if one cluster is invalid hbase-operator#575
• fix: don't stop the reconciliation if one cluster is invalid hdfs-operator#594
• fix: don't stop reconciliation if one cluster is invalid hive-operator#523
• fix: don't stop the reconciliation if one cluster is invalid kafka-operator#773
• fix: invalid objects don't stop the reconciliation spark-k8s-operator#482
• fix: invalid objects don't stop the reconciliation superset-operator#551
• fix: invalid objects don't stop the reconciliation trino-operator#657
• Migrate to DeserializeGuard zookeeper-operator#872
• Migrate to DeserializeGuard commons-operator#293
• Migrate to DeserializeGuard listener-operator#238
• Migrate to DeserializeGuard opa-operator#638
• Secret @nightkr - NOOP, has no K8s controller
• fix: invalid objects don't stop the reconciliation hello-world-operator#127