Updates for (maily electron) FR and scale factors, and other utility stuff

[cippy]

Fixes for scripts about electrons fake rate, now:

1. using all MC samples (W,Z,top,Dibosons, to subtract from data and get both FR and PR)
2. fitting both with pol2 (much better fits than pol1)
3. new FR skim ntuples where produced with all available W,Z and top MC, they are in :
   /eos/cms/store/cmst3/group/wmass/mciprian/TREES_1LEP_80X_V3_FRELSKIM_V8

PR is almost unchanged, while FR decreases by 2% on average (up to 5-10% in EE, but FR was already lower than in EB)
Plots: http://mciprian.web.cern.ch/mciprian/wmass/13TeV/fake-rate/electron/FR_graphs/fakeRate_eta_pt_granular_mT40_35p9fb_signedEta_pt65_subtrAllMC_newTrigSF_fitpol2/

Updating txt files accordingly to get fakes in plots and cards.

NOTE FOR ELE: for the moment, when making cards the FR slope variation syst was removed because we fit with pol2. However, a new shape syst for awayJetPt from 30 to 45 GeV was added, which should be larger than previous syst anyway. This is treated as pdf to make up/down variation, i.e. by symmetrizing with respect to nominal (with same integral)
Ratio of PR and FR: http://mciprian.web.cern.ch/mciprian/wmass/13TeV/fake-rate/electron/ratio_FR_PR/subtrAlllMC_newTrigSF_fitpol2__jetPt30_Over_jetPt45/

Updating scripts and root file managing smoothed scale factors production and dumping them in ntuples. These files are still preliminary so before making new ntuples with SF, let's discuss the status

NEW STUFF:

1. script to make ratios of two existing TH2 in root files: WMass/python/plotter/w-helicity-13TeV/makeRatioTH2.py
2. Script to make a json file running on data, to check if we have lost data when making the ntuples: WMass/python/plotter/w-helicity-13TeV/makeJsonFromData.py

WARNING: I haven't tried yet to make datacards with the new FR systematics, but is should work

[emanueledimarco]

let's keep in mind that we need to introduce it back as a pt-dependent shape uncertainty (@cippy says that this is temporary removed because the FR is now parameterized as a parabola and taking +/- one sigma of the linear coeff doesn't make sense).