Moving special case of keff calculation (n<=3) to cpp side
Created by: matiaslavista
Using test_data: https://anl.box.com/s/u1g3n8iai0u1n5f6ev3pg2j3ff941bqa
From Commit Below: Change in Calculating K-Eff from cpp side for cases of batches (n) <= 3 has changed. Therefore filter_distribcell, which ran batches of 1,1,3 and 1, has been updated to reflect the more accurate calcultion. sourcepoint_batches explanation: This test needed to be updated because statepoint files are written for 3 or fewer batches, which will change the k-eff result written in write_eigenvalue_hdf5 for those statepoint files. All other k-effective values are unaffected and the final result of the simulation is the same.