Fiducial volume coordaintes

[austinschneider]

This PR aims to define the fiducial volume in detector coordinates within the SecondaryBoundedVertexDistribution.

Using geometry coordinates would have caused a weighting inconsistency when weighting with a geometry that has a different definition for detector coordinates than the one used for injection.

This PR also factorizes the logic that loads geometry files so that it can be used to load the fiducial volume, fixes the rotations used in the Geometry <--> Detector coordinate transforms, and fixes some issues with the create() methods.

[nickkamp1]

The code changes all look good here. I'll test that the DN examples still work and give the correct output

[austinschneider]

@nickkamp1 you might want to grab that last commit before you start testing. It looks like I had only reverted changes to the sampling method and hadn't done the same for GenerationProbability etc.

This obviously caused some issues when trying to do weighting in the CCM example

[nickkamp1]

@nickkamp1 you might want to grab that last commit before you start testing. It looks like I had only reverted changes to the sampling method and hadn't done the same for GenerationProbability etc.

This obviously caused some issues when trying to do weighting in the CCM example

Gotcha, tested with the latest commit. There was one remaining issue with the "in_fiducial" flag computed in LIController. This commit calculates that flag consistently with the behavior you implemented in this PR. I think this is good to merge now, will approve