Add aggregate query for datasets

[stuartmcalpine]

Add the aggregate_datasets function to the Query class.

Allow for COUNT and SUM queries in the dataset table.

e.g., datareg.Query.aggregate_datasets("nfiles", agg_func="sum") would sum the total number of files in all datasets.

Rules are the same as general queries, by default both production and working and queried and the result joined. If query_mode != "both" then only the query_mode schema is searched,

[JoanneBogart]

This looks good as it stands, but if you have time it could be enhanced somewhat:

• add support for min, max and avg
• allow queries on at least some other tables (alias, keyword, maybe dataset_keyword) but only if agg_func is "count"
• for anything other than "count", check that the column value type is numeric
• as a convenience for callers, for "count" allow the column argument to be None. The routine should then pick a column or just issue
  select count(*) from the_schema.the_table where...