Using the aggregation Framework using a large set of doc IDs as query? ( + bypassing the scoring part)