This idea comes from ian.boros@mongodb.com's skunkwork project.

We need to evaluate the performance gap in $search and identify the potential improvement achieved by doing ixscan/fetch in chunks. The improvement applies other queries that use ixscan/fetch stages as well.