K8s operator does not correctly scale down a cluster

[shribe]

In 26.4.1

• create a cluster with 3 nodes

• edit the cluster YAM to have 1 replica, kubectl apply...

• the statefulset does get scaled down to 1 node, but it seems to keep returning all 3 hostnames as brokers

[RafalKorepta]

Redpanda don't support scale down plus our webhook prevents from scaling down. Our newer version prevents from scaling down.

REF: https://github.com/vectorizedio/redpanda/blob/0a7d76359befb7763ca7274543ca5ca1c69504b5/src/go/k8s/apis/redpanda/v1alpha1/cluster_webhook.go#L88-L93

[shribe]

OK, I did this via kubectl apply..., in version 21.6.4.

[emaxerrno]

1591 and #1654 and #1710 and #1695 allows us to scale down next. most of them are merged, so @RafalKorepta this is a valid tix.