Detector hit ref bug

[StackEnjoyer]

Currently, the Detector stores hits by directly referencing the underlying AbstractBeam or AbstractRay type. In Julia, this is equal to reference via a pointer. The issue is that for cases where a beam is retraced onto the detector several times without intending to use empty!, multiple hits are stored which reference the most recent state of the beam, making this use case impossible.

Possible fix: some form of deepcopy when generating an AbstractHit in addition to a testset.

[StackEnjoyer]

It appears that deepcopy is not the desired quick fix, since some circular references exist within the general BMO architecture. This means that custom, ref-free hit types are required for a "simple" solution, since deepcopy grinds to a halt for multiple subsequent calls, e.g. on the GaussianBeamlet.