SERVER-112691 added long/short running metrics for the execution stats similar to the one already available for normal/low priority stats. However, we might want to reconsider the structure of some execution metrics. The issue is that longRunning/shortRunning (and normal/low priority) is a per-operation designation, not a per read/write lock acquisition one (at least according to the current heuristic). For most metrics this is fine enough, but it doesn't work for newAdmissions. Every operation begins as a short running operation and only becomes long running after a certain number of acquisitions, but we only increment newAdmissions when the number of admissions is 0, so longRunning.newAdmissions (and normalPriority.newAdmissions) never gets incremented.

One idea would be to consider flipping the metric hierarchy to have longRunning and shortRunning being at the top level, and then we could have a new metric called totalOps that are incremented when an operation terminates. Or, if we keep it under read/write, it would count the number of operations that end as read/write.