mesh tally bug with time census

[northroj]

I was adding a time census version of the inf_shem361 problem in our regression suite for testing some of the cell tally changes I made, but the mesh tally in the problem errors when census is enabled. The stdev in the final energy group of the mesh tally is a NAN. The mean looks reasonable.

error printed at the end of the final census (it looks like this code happens as tally post-processing): MCDC\mcdc\kernel.py:2340: RuntimeWarning: invalid value encountered in sqrt tally[TALLY_SUM_SQ] = np.sqrt(