Scope doc for collecting query stats for errors.

Need to figure out the downstream impacts + feasibility of either renaming execCount. Right now, execCount records the number of successful queries. Query stats for errors work introduces an execCountErrored metric, so "execCount" may be a misleading name for users. If unfeasible, it may be worth introducing a new metric to signify total error count (successful + errored).

Working on SERVER-131926 revealed another reason we may want to think about updating semantics or including new metrics: right now, when a cursor fails on batch N, metrics for batches 1-(N-1) are still stored in query stats store and execCount is bumped. However, introducing execCountErrored leads to this count being incremented too. This could be unintuitive for users.