Product would like to answer the following questions about $vectorSearch usage: 

• What types of MQL style filters are users using and with what data types?
• How many vectors are being requested on average per index (limit)? 
• What value do users put for numCandidates?

Right now the $vectorSearch query shape is a black box, it doesn't report the values of any specific fields. We will capture the above information by doing the following:

• If present, include filter and index in the query shape for $vectorSearch. This will allow stats to be aggregated by index if desired.
• Add a new SupplementalStatsEntry implementation for vector search. This will include metrics reporting the limit value as well as the numCandidatesLimitRatio.