new projection code for MELooper: sparsinator.py

[jpata]

the MELooper.cc code is needlessly complicated. I've a new candidate for creating the sparse histograms used for downstream limit setting & systematic data/MC comparisons in

https://github.com/jpata/tthbb13/blob/meanalysis-80x/Plotting/python/joosep/sparsinator.py

It replaces the MELooper.cc and can be extended to produce additional non-sparse histograms. It still needs to be extended for JER systematics and re-weighting systs.

[jpata]

This code could already be tested on

• MC: https://github.com/jpata/tthbb13/tree/meanalysis-80x/MEAnalysis/gc/datasets/Jul15_leptonic_v1
• Data: https://github.com/jpata/tthbb13/tree/meanalysis-80x/MEAnalysis/gc/datasets/Jul18_data_v1

[jpata]

the new sparsinator no longer auto-splits the joint tt+other samples and no longer uses the short process names (ttH_hbb). Instead, it creates directory structures like

sample/sl/sparse{_syst}, where sample is the full name of a sample such as TT_TuneCUETP8M1_13TeV-powheg-pythia8. This is so that histograms from different tt+jets samples could easily be merged.

The sample can be split based on that name and the process axis of the sparse histo.

[jpata]

MakeCategory.py is working again.