Add uniform field + MSC along-step kernel

[sethrj]

This implements an alternate along-step kernel for a uniform magnetic field with optional MSC. To be consistent with adept's default, this kernel doesn't implement energy fluctuations (though it would be trivial to add that capability).

Closes #281 .

[amandalund]

Nice, @sethrj! Just a couple build errors and conflicts to resolve.

Were you able to run the demo loop with a uniform field?

[sethrj]

Running TestEM3 with a 1T field along x gives an error:

/home/s3j/.local/src/celeritas/src/celeritas/global/alongstep/AlongStep.hh:120: void celeritas::along_step(MH &&, MP &&, AE &&, const celeritas::CoreTrackView &) [with MH = celeritas::detail::UrbanMsc; MP = lambda [](const celeritas::ParticleTrackView &, celeritas::OrangeTrackView *)->celeritas::FieldPropagator<celeritas::FieldDriver<celeritas::DormandPrinceStepper<celeritas::MagFieldEquation<celeritas::UniformField>>>>; AE = celeritas::detail::EnergyLossApplier]: block: [7837,0,0], thread: [68,0,0] Assertion `mfp > 0` failed.

😢 I'll see what might be going on. This is the first time we've ever tried to run with the field propagator outside of unit tests...

[amandalund]

Haha yep I just tried the same and see lots of those assertion failures (boundary action with zero mfp, all particle types)...

[sethrj]

@amandalund Somehow the field propagator is moving all the way to the end of the physics step but marking it as a boundary crossing. I think this is a bug in the propagator as opposed to the new code: the propagator tests are far from complete. I suggest we merge this branch (knowing that it doesn't work in anything but an infinite medium) and I rewrite the field propagator tests and fix the propagator.

[amandalund]

Works for me. Thanks @sethrj!

[sethrj]

CI failure is due to out-of-disk for the vecgeom demos as usual.