Logstash resources are not reconciled when upgrading from ECK 2.11.x to 2.12.0-SNAPSHOT

[barkbay]

How to reproduce

1. Deploy ECK 2.11.1
2. Apply the sample resource config/samples/logstash/logstash_es.yaml
3. Wait for the Logstash Pod to be created
4. Upgrade ECK to main

The following log message is printed endlessly:

2024-02-09T09:10:46.459+0100    DEBUG   logstash-controller     Some pods still need to be created/deleted      {"service.version": "0.0.0-SNAPSHOT+00000000", "iteration": "12", "namespace": "default", "logstash_name": "logstash-sample", "namespace": "default", "statefulset_name": "logstash-sample-ls", "pending_creations": ["logstash-sample-ls-0"], "pending_deletions": []}

The root cause is that expectations for Logstash are never considered as "met", it's because label logstash.k8s.elastic.co/statefulset-name is originally missing on Pods created with ECK 2.11.1:

> k get pods
NAME                                READY   STATUS    RESTARTS   AGE
elasticsearch-sample-es-default-0   1/1     Running   0          8m22s
elasticsearch-sample-es-default-1   1/1     Running   0          8m22s
logstash-sample-ls-0                1/1     Running   0          8m21s ### Created with ECK 2.11.1

> k get pods -l logstash.k8s.elastic.co/statefulset-name=logstash-sample-ls
No resources found in default namespace.

Therefore existing Pods are never returned by GetActualPodsForStatefulSet.