PGeant model not yet included in tests against Geant data

[GilesStrong]

PGeant model doesn't model the bulk of scatterings as well as the PDG model, and is currently excluded from the integration tests comparing the geant distributions to the scatter models (tests require a 2sample Kolmagorov Smirnov test with p-value greater than 0.01 in the 68th percentile of the absolute value of the distribution).
Either we improve the PGeant model, or relax the tests given that the PGeant model is expected to better model the tails of the scatter distributions.
See also issue #132

[giamman]

At least ad interim, a way to relax the tests could be to restrict the comparison to the bulk. Does that sound meaningful?

[GilesStrong]

@giamman Yes, that's currently what I have set up: The tests only check that the bulk is well modelled https://github.com/GilesStrong/mode_muon_tomography/blob/main/tests/test_scattering.py#L228-L248, although KS tests are computed on also the full range and 95th percentile range.
Even so, the PGeant doesn't model the bulk well enough to pass the checks.