Separate the data histograms/events from the Propagator object

[nadrino]

The current implementation of the Propagator has been made on the assumption that we would always use it in order to compare a "Data" to a "MC" histogram.

While writing the CalcXsec tool, I realised the feilds of application of the Propagator would be wider. For instance, the CalcXsec never uses the "Data" slot since in only needs to propagate a set of parameters on the MC samples.

A more suitable stucture for the future would be to redefine the Propagator as a class structure that propagates a set of parameters to a given set of samples, i.e. only MC. The data histograms might be built using a separate Propagator object.