Add ability to perform a restoration without downtime

[c3-clement]

What is missing?

When restoring a Cassandra cluster, the operator will first fully stop the Cassandra statefulset and then re-start it in restore mode.
It would be great to be able to perform a restoration without bring the entire cluster down.

Why do we need it?

In my company, we want to implement multi-tenancy environments. In the Cassandra layer, each tenant would be represented by a keyspace.
Therefore, we need to be able to restore a single keyspace, without downtime on other tenants.

Anything else we need to know?:

This issue is closely related to #1445

┆Issue is synchronized with this Jira Story by Unito
┆Issue Number: K8OP-288