What is missing?
Improved handling of rule configmaps in a grow-only mode.

Why do we need it?
When reaching the size-limit of the first configmap and then falling below it again, the operator updates the STS and recreates the pods to go back to 1 configmap. This introduces disturbance - especially if the number of rules is around the limit, therefore re-triggering this behaviour multiple times.

I would propose just leaving the new ConfigMap empty in this case and not removing it, therefore reducing the impact on the running Prometheus.