For the CBR for no MP results fallback mode, in the case that the initial MP trials find some docs (but not enough to EOF or fill a batch), we resume multiplanning by calling runTrials() a second time. This means that in the initial trials we increment the metrics once, and when we resume we increment them a second time, breaking timing information up into two parts and causing duplicate counting of MP invocations.

The metrics affected by this are:

• query.multiPlanner.histograms.classicMicros
• 
  query.multiPlanner.histograms.classicWorks
• query.multiPlanner.histograms.classicNumPlans
• query.multiPlanner.classicCount

server_status_automaticCE.js should also be updated to test this behavior, as it currently uses a query which hits EOF to resume MP rather than finding just a few documents.