Normalisation by vanadium needs to take bin widths into account

[jl-wynen]

Similarly to #167, the normalisation by vanadium also needs to take into account the bin widths of its inputs (sample and vanadium run). Especially so because we may pre-process vanadium and save it, possibly is a histogram which likely won't match the bin width of the sample run.

This is a problem when normalising in event mode. At that point, we know the bin width of the denominator but the numerator's bin width can still change before it is finally histogrammed. So we may need to histogram before normalisation.

[jokasimr]

If we make sure to histogram the saved vanadium run to a relatively dense grid we should be able to rebin that to the coarser grid of the sample run without distorting the normalization factor much. Is that a feasible strategy?