Update cluster-wide SQL

Picodata' SQL relies on the sbroad library to execute clusterwide queries. Sbroad

1. Was refactored in the way it calculates buckets for the INSERT queries. All these changes allowed us to use a hash dunction from the key_def to calculate bucket_ids in Picodata.
2. Has lot's of bug fixes and improvements

So, current MR updates sbroad version, improves its documentation and adds some tests to check closed bugs in the Picodata SQL runtime.