fix: add check to validate cp node name = token declared node-name

[ethandcosta]

Currently, microcluster does not validate that the name of a joining node matches the name attached to the token in joins with. So for example, I can do get-join-token on node x, and use it to join with node y. The docs [doc/clustering.md] says that this behavior should not be the case.

This causes problems for us in k8s-snap, where we need to enforce node name deduplication to prevent data store drifts between microcluster and etcd (canonical/cluster-api-k8s#229) and indeed we are doing this for worker nodes, but we cannot affect this behavior for CP nodes, as their lifecycle is owned by microcluster. This fix should prevent joins where a node uses a token that declares a different node name

[roosterfish]

#204 added the ability to account for member name changes after creating the token.

In this PR, this commit added the check you see a few lines above. If the SAN field of the cert provided by the joiner does not contain the name of the join token record, it errors out.

So my question is why do you require this extra check, if the one above should already error out? When the new member creates its server.crt, it uses its own name.
Do you maybe modify the members server.crt prior to performing the join?