Fix photon tally issues when using analog/collision estimators
Created by: liangjg
This fix #1207 (closed) .
The problem is, whenever a collision is done, we check the energy of the particle and the particle weight (actually
wgt_last_) is set as zero if the energy is below cutoff energy. However, this is done before the analog/collision tally being scored, causing zero score for the last collision events. This didn't occur in neutron case because the neurton cutoff energy is usually zero and this never happened. Of course this doesn't affect tracklength tally as it is done prior to the collisions.
Another bug fixed in this PR is the segment fault which is about the wrong index of particle filters.
The tests are updated to cover these two bugs (the model is enlarged to make sure collision is able to occur).