Performance improvement proposal for mongoose Query streams

I've benchmarked mongoose streams against mongodb streams and found mongoose was one order of magnitude slower.

I think there is a performance problem on current implementation, relying on continuously using mongo's cursor.next along with non-trivial flow control code to mongoosify returned documents.

IMHO, using an streaming approach involving mongo's Cursor.pipe along with stream's Transform, thus leveraging the flow control entirely to the efficient Readable.pipe interface, will result on much better numbers and, maybe, simpler code.

No files or tests are named. Start by reproducing the benchmark comparing Mongoose and MongoDB streams, then inspect the current cursor.next flow-control and document-conversion path; done means validating whether a Cursor.pipe-based approach improves the reported performance without losing stream behavior.

Written by the indexing model from the issue text.