Extensible threshold and likelihood computation

Allow for different likelihood setup than the current RSSE w/ normally distributed residuals.

This would mean allowing to create "backends" for the calcViabilityThreshold and calcNegLogLikelihood most likely via options (handle or string name of the function to call, with the pre-defined interface). Consequently, because input data (IQMmeasurement) has {min,max}values concept, the ExpData class has to be generalised a bit from the diagonal covariance matrix-specific solution, which currently interprets difference between these two extreme values as two std deviations.

It would make sense then to rename the functions to a more informative names as: calcDataViabilityThreshold and calcDataSimulationError.