Including background systematic uncertainties in gridscanner

Hi @bejaeger

Following on from !709 (merged), it would be nice if there was a way to include the effect of background systematic uncertainties on the poisson (with error) significance definition. Currently, I don't think there is a way to do this.

I'm happy to go about trying to implement this, but wanted to discuss here initially.

Our analysis (0nuBB) has not evaluated a complete set of systematic uncertainties, but we are expecting a ~15-20% relative uncertainty on the background. Therefore, for now, just a simple flat normalisation uncertainty would be sufficient, and not something more complex with the band files.

My suggestion is to add a tag in the gridscan config file (default 0):

inclRelBkgNormUnc: 0.15

and then in here one could have something like

auto bAndErr = bkgAndError(iregion); //  pair of bkg yield and uncertainty
b = bAndErr.first;
berr = bAndErr.second;
const double relativeBkgErr = getTagDoubleDefault("inclRelBkgNormUnc", 0);
berr = TMath::Sqrt( TMath::Power(berr, 2) + TMath::Power(relativeBkgErr * b, 2) );

what do you think?