Time-of-flight modeling
We need to think about how much time has passed between to discrete processes. This is related to how trajectories are parametrized (r(t), r(u), r(X), ...). Due to energy losses, speed is not constant but the trajectory in our implementation does not know about this (of course, this makes a difference only for low gamma factors). Furthermore, there is Konrad's demur that the "real" trajectory is in fact longer due to multiple scattering, which should cause a delay compared to the straight line.