Create one or more new metrics related to the ClusterCatalog cache

[perdasilva]

Follow up that came out of #1207

Should we create one or more new metrics related to this cache? For example:

- gauge: number of known ClusterCatalogs
- gauge: number of local caches

Or maybe just store the diff as a single gauge (that we would expect to always be 0)?

Originally posted by @joelanford in #1207 (comment)