Allow for aggregate queries in Python MongoDBIO Read

[damccorm]

Currently, MongoDBIO in the Python SDK only allows to use basic projections and filters (db.x.find(..)), but not aggregations (db.x.aggregate(..)).

As a pipeline creator, I would like to be able to use aggregations and more complicated queries as input to a beam pipeline. Example use cases are sampling data ($sample), joining foreign data ($lookup), 

Imported from Jira BEAM-8887. Original Jira may contain additional context.
Reported by: corvin.