Prevent or handle replacing unbootstrapped nodes

[mflendrich]

Issue originally authored by tnozicka as #855

If a user or the AutomaticOrphanedNodeCleanup feature replaces a node that isn't bootstrapped yet, we get stuck with:

ERROR 2021-11-25 13:42:27,194 [shard 0] init - Bad configuration: replace-address and replace-address-first-boot are not allowed for seed nodes

The issue is that the operator doesn't know if scyllacluster is bootstrapped or not. We need to propagate the information into an API object so the operator can distinguish the case and generate a valid config.

There are also other places where we act based on readiness while we need to assess only if it's bootstrapped. Given that we don't correctly handle cases where the node is not ready, yet it's bootstrapped.