kv: include all replicated locks in `crdb_internal.cluster_locks`

As discussed in the [20220104_cluster_locks.md RFC](https://github.com/cockroachdb/cockroach/blob/42a22e69788a22851fc09fc57a20dcbf63d00a75/docs/RFCS/20220104_cluster_locks.md), we don't currently pull replicated locks into memory and include them in the table if they're not already contended and resident in the partial, in-memory lock table. This makes the table confusing and difficult to use.

We should change this, at the cost of making queries to this table more expensive. These queries should be properly paginated and paced such that they don't risk impacting performance, even if they do hit disk.

Jira issue: CRDB-32751