Hadronic model interface overhaul

There must be a more clean-cut separation between interaction models, and the cascade-internal managements. In particular, the interaction models should NOT know anything about the (complicated) C8 Stack and the ProcessSequence and in also the Environment.

This also means, quite some redundant code currently scattered through many interaction models must be centralized inside ProcessSequence and Cascade. E.h. sampling the target nucleus cannot be done in each interaction model... etc.