branching_ratio not taken into account in normalisation

[KrisThielemans]

There's a bug in BinNormalisationWithCalibration. Its get_branching_ratio uses an uninitialised radionuclide member, therefore giving branching_ratio=1 (and writing a warning).

There's a fix on https://github.com/KrisThielemans/STIR-1/tree/NormWithCalibrationFix. A PR will follow, but it's probably best to do that together with other changes to the norm (and allow for disabling it).