Proper application of b-tagging SF

[ktht]

As argued in this thread on CMS Talk, the application of shape-changing b-tagging SF implies that we first obtain the distribution of AK4 jet multiplicity with and without the b-tagging SF plus no cuts on b-jet multiplicity in the SR of a particular analysis, and then apply the ratio of the latter to the former as a function of jet multiplicity in order to obtain the correct estimate of the MC process in that analysis phase space. For us, it implies the following:

• in flat Ntuple production, store nominal event weight with (already present) and without b-tagging SF;
• at the analysis level (ie in this repository) we need two features:
  • ability to save the jet multiplicity distributions in a histogram for the central case as well as for the b-tagging systematics;
  • ability to apply the ratio of the two histograms as an additional event weight at the analysis level;
• regarding the analysis workflow, the ratios could be obtained automatically (ideal case) or computed only once (in case it's too expensive to run). In the latter case it would be the responsibility of the user to rerun this part of the workflow every time the b-jet selection, b-tagging discriminant, b-tagging SF or SR cuts change.