OF-2502: Do not iterate over clustered cache when calculating stats
When collecting statistics, two MUC methods iterate over caches. That is inefficient in itself, but the cache can also be clustered. If that's the case, then a lot of data serialization, which is a waste of resources.
This commit introduces a new cache in which statistics can be stored. This way, the statistics need not be recalculated every time, saving resources at the expense of potential inaccuracy.