Background

Currently, the server populates `batchSize` in CursorOptions for SearchCommand and GetMore for queries initiated by `$search`. If it is not populated, Mongot will compute scores and read _id for 101 top documents. 

Problem:

`searchMeta` is used for pure facets results or count, and Mongot doesn't need to do scoring or _id lookup. However, Mongot currently has no indication it's serving a $searchMeta request, so it still retrieves and scores 101 results. Additionally, we have work scheduled for embedded facets where we can improve future user experience for searchMeta queries by accepting simplified syntax.

Solution

Mongod should indicate to Mongot that it is performing a `$searchMeta` query via a new boolean parameter isMeta/requiresDocs in CursorOptions. This solves both issues: Mongot can accept simplified syntax for embedded facets, but also we can improve performance on existing searchMeta queries by internally running a batchSize of zero (which ordinarily only valid for establishing sharded cursors).