Reconsider transverse diffusion treatment at the interface

[ramirezdiego]

Right now we smear our S2 pattern map according to transverse diffusion, via s2_pattern_map_diffuse, where we redefine our S2 pattern for it to account for this diffussion. This means that we treat the electron cloud at the interface as a point, apply the extraction efficiency for that XY, and later on effectively correct for diffusion at the PMT pattern level.

The alternative (more physically correct approach) is to get rid of this function and apply the transverse diffusion at the interface, such that the extraction efficiency per electron is better applied and we do not need corrections for the pattern map.

[ramirezdiego]

(The bad side of this implementation is that we apply a per-electron treatment earlier in the pipeline, which impacts both memory and speed of the code.)