Query and Secondary index support for non-JSON

This is a potentially umbrella improvement for understanding the desired behavior, implementing, documenting, etc everything around allowing users to work with binary data through N1QL and 2i.

Specific attention should be given to the common use of counters and strings within the k/v use cases of Couchbase Server as well as secondary indexing for range scans on the key names itself while the document value may be un-parseable