Development of altBkg fits with MC templates

[rubenforti]

This PR contains some changes that allow to run the alternative background fits using a template extracted from MC predictions. Below are the most important changes.

Import and use of the templates

• Defined sample for the bkg histograms in tnpEGM.py
• Created the function histFitterAltBkgTemplate in libPython/fitUtils.py to handle the specific options for the fit (it replaces the histFitterAltBkg function in tnpEGM.py). The Barlow-Beeston strategy can be activated by turning into True the argument isBBfail of this function
• Introduced methods setTotalBkgShapes and setBarlowBeestonBkgPdf in the tnpFitter

Technical adjustments

• Some changes in tnpFitter::manageFit to run on ROOT 6.32 and to avoid parallelization of the minimization procedure for the single bins (see release note): without that correction, some instabilities in minimization convergence have been observed.
• The CMSShape pdf now is defined with an analytical integral. It has been seen that, without this feature, when running on the 'tracking' step, the minimization procedure wasn't able to end for some bins. This effect has been observed before the implementation of the correction explained previous point

Other adjustments

• New file create_totBkg_file.py to produce the TH3D representative of the sum of the various bkg sources (scaled wrt data luminosity)
• The "main efficiency plots" produced by the standard procedure, those containing three types of efficiencies, now show the nominal fit together with the altSig and the altBkg ones. The fit on MC is not shown anymore, but the MC counting efficiency is reported below the nominal one