Instead of updating the values in CurOp every time memory usage changes, stages that use the chunked memory tracker factory methods only update memory usage to CurOp once their memory usage has changed by a certain chunk size. 

We could update CurOp either at the end of each getMore (so numbers are at least more accurate between requests). If this slows perf down too much, we could also just update CurOp when we get a final number at the end of the stage's execution, much like we do for updating peakTrackedMemBytes for explain execution stats.