k8ssandra-operator shouldn't modify the cassdc objects on operator upgrade

[adejanovski]

k8ssandra-operator in spec changes to the cassdc objects which will instantly trigger a rolling restart when the operator is upgraded.
As one operator instance may manage lots of clusters, we don't want upgrades to trigger a rolling restart storm.

Tasks

Beta Give feedback

1. If the Generation of the K8ssandraCluster is the same as the ObservedGeneration we don't update the CassDC objects in Kubernetes (instead, we mark the K8ssandraCluster as requiring refresh). Metadata updates would still be applied.
2. A condition should be placed on the object to indicate that a restart is pending (something as "RestartRequired" condition)
3. Restart can rely on a new annotation k8ssandra.io/allow-upgrade=once (using k8ssandra.io/allow-upgrade=always would allow permanently to automatically update the K8ssandraCluster on upgrade). This annotation would be placed by either the users or the rolling restart K8ssandraTask. If this annotation is detected, the CassDc Spec update will be unblocked, the "RestartRequired" condition will be set to false and the annotation will be removed at the end of the reconcile if its value is set to once
4. We need to allow reconcile to trigger if this annotation is placed on the objects (k8c). This requires an additional predicate for annotation listeners, but then one more filter. This could be useful outside of this feature also.