Creation of (un)conditional snapshots in Impact and Breakdown calcuators seem inconsistent

It has been observed that the minimization of the NLL (when trying to create a snapshot) in the WriteCalculateBreakdown Action (the conditional and unconditional pre-fits) can fail while it will then just happily converge in the WriteCalculateImpact Action.

In the first case the NLL evaluates to >1e30 (considered as 'inf') while in the second case it shows moderate values during minimization (O(1e3)-O(1e4)). It seems like there might be some inconsistency between the two prefit implementations

TODO: check, unify and remove code duplication if applicable.

relates to #82 (closed) as these two issues were first observed in conjunction