Adding option to scale RMS uncertainty

[dittmer]

Adding the option to rescale a RMS uncertainty (ex. 2016 PDF) during the nuisance postprocessing
This allows ex. the normalization component of these uncertainties to be dropped
Tested in WW differential analysis -- see https://github.com/latinos/PlotsConfigurations/blob/master/Configurations/WW/FullRunII/Full2016_v7/inclusive/nuisances.py#L507
Syntax is to add a line with 'scale' : { sample : [scaleup, scaledown], ...} to nuisance

[amassiro]

Just a suggestion.

This is another point where you can achieve the same result with minimum changes:
https://github.com/latinos/LatinoAnalysis/blob/master/ShapeAnalysis/scripts/mkDatacards.py#L325
and no need to run mkShape.

The idea is that if in the datacards we have
shape 1.0
it means that the up/down histograms correspond to +/- 1 sigma.

If we want to double the uncertainty, we can write
shape 0.5
it means that the up/down histograms correspond to +/- 0.5 sigma, thus effectively doubling the uncertainty.

If then post-fit plots are shown, this, I hope, will be taken into account by combine