Cluster replication should support permissions and documentation for such on `dolt_assume_cluster_role`

`CALL dolt_assume_cluster_role` is very powerful and can be used to disrupt the availability and in some cases the committed writes of an existing doltdb cluster. We should provide permissions around it at the SQL layer and guidance around how to control access to it.

Start by locating the SQL entry point for CALL dolt_assume_cluster_role and the existing permission-checking and cluster-replication documentation. Determine how access is currently controlled, then define permission behavior and add guidance for limiting access. Done means unauthorized calls are blocked as intended and the documentation explains the security and availability risks.

Written by the indexing model from the issue text.