Stochastic photon propagation

PROPOSAL differentiates between continuous and stochastic energy losses. This is necessary for charged particles like electrons, since the bremsstrahlung crosssection is diverging for small energy losses, which means that we have to define a part of the crosssection that is treated continuously.

At the moment, this treatment is also done for photons. Here, all small Compton energy losses are treated continuously. However, this is not necessary for photons, since there is no diverging crosssection. Using very-small cuts, I experienced that the continuous treatment of Compton scattering becomes numerically unstable. And in general, having continuous losses for photons (albeit very small) doesn't feel right 😅

Therefore, this PR introduces the changed to the interface to use an only-stochastic treatment of photons instead. For the simulation results and the runtime of the simulations, this change seems to be negligible.

I've made a test for 1000 electromagnetic showers with an energy of 1 TeV and a cut setting of 2 MeV:

Runtimes remain almost unchanged as well (improvement of about 3%, but probably not statistically significant).

We might want to test this with more simulations, maybe also radio simulations.