added negative effciency protection

I've seen some LBs reported with negative efficiencies. This is due to the computation of the efficiency which summs up all the busy time in an LB, divides it by the total LB time and, because this is the efficiency of taking data (not of being busy), it subtracts the ratio from 1. But adding all the busy sources in an LB, especially with many busy detectors, can give a total busy time > total LB time...and thus <0 efficiency after the subtraction.

The fix was that when the total busy time goes over the LB time, the efficiency reported should be 0.