Make enable_pod_antiaffinity configurable

[eberlep]

For cases when it is unfavourable to have more than one pod of the same postgres cluster on the same worker node, the ability to enable pod anti affinity (which is already exists as a built-in feature of the postgres-operator) should be implemented. Once enabled, only one pod of the same postgres cluster will be scheduled on the same node at a time.

[eberlep]

When enabling the setting in the postgreslet configmap, it is handed down to the operator itself:

 ~ $ k get cm postgres-operator -o jsonpath="{ .data.enable_pod_antiaffinity }"
true

Now when privisioning a 3-replica postgres cluster in a single kubernetes cluster, the scheuler (expectetly) refuses to start more than one pod:

 ~  $ k get po -o wide
NAME                                 READY   STATUS    RESTARTS   AGE     IP             NODE                                           NOMINATED NODE   READINESS GATES
pgfits-podantiaffinity3d002-0        3/3     Running   0          2m49s   10.244.0.208   shoot--abcdef--pgaas-dev-group-0-79985-t6zvn   <none>           <none>
pgfits-podantiaffinity3d002-1        0/3     Pending   0          2m43s   <none>         <none>                                         <none>           <none>
postgres-operator-7f5b947b89-fb6zd   1/1     Running   0          2m59s   10.244.0.206   shoot--abcdef--pgaas-dev-group-0-79985-t6zvn   <none>           <none>