Remove Cluster label in TiMonitor's prometheus scrape job.

Backgroud

In my multi-kubernetes monitoring setup, we are using cluster label as external_labels to indicate an unique kubernetes cluster(I guess there are some more people doing the same🤔), when deploying timonitor and send its metrics to our central monitoring storage, I find some confusing results, cluster label are setted as tidb cluster.

Information I found

After digging into the impl, I find there are two kinds of cluster label providing almost the same information:

1. cluster label:
   https://github.com/pingcap/tidb-operator/blob/d4c238c3d36377487f2225b627fc38a429e5c219/pkg/monitor/monitor/template.go#L375-L380

2. tidb_cluster label
   https://github.com/pingcap/tidb-operator/blob/d4c238c3d36377487f2225b627fc38a429e5c219/pkg/monitor/monitor/template.go#L396-L402

Thus, I think it is okay to leave only one of them, namely tidb_cluster. This way, also avoiding the conflict I mentioned above.

Suggestions

I think we can first provide a config to disable this relabel setting for the compatiblity, and fianlly remove it after few versions later when we community reach concensus.

Start in pkg/monitor/monitor/template.go at the linked sections around lines 375-380 and 396-402, and compare how the two cluster labels are added to the Prometheus scrape job. Review the compatibility suggestion before deciding the configuration scope. Done means avoiding the conflicting cluster label while retaining tidb_cluster, with the transition behavior agreed by the community.