[CX] Perf: Determine typical filter conditions

Analytics allows the creation of zero or more datasets for each bucket. Each dataset contains a filtered subset of the data in the bucket. We need to determine what typical filters for this ingestion to include those in performance testing.
Specifically the performance testing should cover recommended approaches to modeling sub-collections in a bucket in Couchbase like e.g. the content of a "type" field or a common key prefix for the keys in the KV nodes. When collections are available in the KV nodes, they should be used for filtering as well.