Propagation Code: Refactoring, Stage 1

This is a first step in refactoring the propagation code to be come more modular. After a report of some very interesting findings by @cscharf and @bbrueers this also includes a change in the order of processing transport contraints, and hence some breaking changes:

• The diffusion is now calculated based on the pre-step e-field and oping concentraiton values instead of the post-step. While for most situations this change is minor, they see large differences since they operate with laser-generated charge carriers close to the sensor surface. Steps outside the sensor volume, even if recovered, hence always lead to zero diffusion in the past.
• Corrections for the final position (state = HALTED but also surface interpolations) should happen before recombination, trapping and gain are calculated.
• Gain and trapping are only relevant if not already HALTED.

More changes to come in a follow-up MR, which also attempts to make the lookup of field and doping more efficient by reducing the number of calls.