Operator should prevent reducing the number of cores

[nicolaferraro]

Version & Environment

Redpanda version: (use rpk version): latest

What went wrong?

If a user creates a Redpanda Cluster CR with e.g. 3 cores (computed from a combination of CPU/memory request) and then reduces it to e.g. 2 cores, the operator currently allows the change but the node fails to start.

What should have happened instead?

The operator should block changes resulting in a reduction of cores, to prevent node downtime. Possibly it should enforce this only for already running clusters.

How to reproduce the issue?

1. Create a cluster CR with spec->replicas=1 and spec->resources->request and limit set to cpu=3, memory=6Gi
2. Wait for the cluster to run
3. Update the CR setting spec->resources->request and limit to cpu=2, memory=6Gi
4. The pod will crash on start

cc: @alenkacz , @0xdiba