Add prometheus label "raft_cluster" to non-QQ raft metrics

[the-mikedavis]

Adds a <<"raft_cluster">> tag for ra_metrics emitted the stream coordinator and Khepri. By default Ra sets the cluster name as the metrics_key (if not set in config, like it is for QQs). The prometheus plugin ignores atoms that aren't protocol names when rendering labels so these cluster name values disappear. We can convert them into a map #{<<"raft_cluster">> => atom_to_binary(ClusterName, utf8)} to render properly.

Before:

$ curl -s "localhost:15692/metrics/detailed?family=ra_metrics" | grep rabbitmq_raft_term_total
# TYPE rabbitmq_raft_term_total counter
# HELP rabbitmq_raft_term_total Current Raft term number
rabbitmq_raft_term_total{vhost="/",queue="qq"} 9
rabbitmq_raft_term_total 10

After:

$ curl -s "localhost:15692/metrics/detailed?family=ra_metrics" | grep rabbitmq_raft_term_total
# TYPE rabbitmq_raft_term_total counter
# HELP rabbitmq_raft_term_total Current Raft term number
rabbitmq_raft_term_total{vhost="/",queue="qq"} 9
rabbitmq_raft_term_total{raft_cluster="rabbitmq_metadata"} 10

Connects #9665
Connects #12142

[the-mikedavis]

This doesn't quite solve #12142 but fixes the issue of these Raft metrics showing up without a label for Khepri and the stream coordinator. There are some larger changes @kjnilsson and I have been looking at for seshat, ra and here to improve how we expose Raft metrics: switching from the ra_metrics table to Ra counters, improving seshat's usability for Prometheus. In the meantime though this is a nearly trivial fix that we can make and easily backport so I will merge this down as-is and leave the rest for future work.