The field OpDebug::workingTimeMillis causes some confusion because it's so similarly named to OpDebug::AdditiveMetrics::clusterWorkingTime.

Most (all?) of the metrics in AdditiveMetrics are used for reporting query stats.

workingTimeMillis is used for slow query log lines, and only represents local time to the current process. It is not used for query stats at all. We should rename it to something like localWorkingTimeMillis and call out in a comment that it is completely separate from query stats metrics.