Associate the collection counts by collection ID instead of by name

If a collection is dropped and recreated with same name the collection reflects the same number of documents it had prior to dropping the collection.

Thanks to Dave Finlay for the following explaination:

For IDs of things (like buckets) in Prometheus, we store just the names.

Therefore, kv_curr_items, for example, only records the item count by bucket name so if a bucket is deleted and recreated with the same name it's going to take up to 20 s for the UI to realize that the bucket item count has abruptly dropped to zero.

However, we do associate the IDs of collections with the metric information. Therefore the UI should be changed to associate the collection counts by collection ID instead of by name.