Measurement fixes

[senkin]

A few fixes and improvements in the cross section measurement workflow:

• Updated configuration to work with the latest (AN-14-071_3rd_draft) set of histograms;
• Apparently we were still changing both fit results input and the response matrices for ttbar generator systematics. Fixed that (now just changing the response matrix);
• Implemented the proper calculation of the hadronisation systematic (difference between powheg_herwig and powheg_pythia). Powheg_herwig is not available for 7 TeV in 53X and not sure if it's in production. Alternatively, can use the old one from Fall11, 42X (/TT_AUET2_7TeV-powheg-herwig/Fall11-PU_S6_START42_V14B-v1/AODSIM);
• Added the top pt reweighting MadGraph (fixes Add MadGraph + top pT re-weighting to results #88) in the cross section plots, the truth histogram is from the unfolding workflow (not fit results like before). The corresponding systematic is also calculated correctly;
• --ptreweight flag in the 02_unfold_and_measure.py script switches the central measurement to be unfolded with pt-reweighted MadGraph. Not used by default (let the ARC decide);
• Fixed PDF calculation (using a different response matrix for each PDF variation). The PDFs are currently preliminary (x/y axes swapped and some are missing);
• Commented out pt reweight systematic;
• Tidied up tables, made compatible with the AN tdr
• Fixed the bug in calculation of the rate-changing systematics (issue Rate-changing systematics are all zeros #110)

[kreczko]

I assume this is now complete, is it?

[kreczko]

I am not a fan of code in comments. Could you please remove it if it is not needed any more?

[senkin]

Shouldn't we keep it until after the ARC review, in case it's decided to change both the input fit results and the reponse matrix for the generator systematics?

[kreczko]

Then we can always got back in history (git time machine ;) )

[senkin]

But then we'd encounter a git paradox!
Fixed.

[senkin]

This can be merged (but please go through the code changes). In case if there's a strong need to implement categorisation of the systematics in the tables (rate/shape-changing), I can do it now, otherwise can leave it for later when we start fixing cosmetics.

[kreczko]

I don't think we need the categorisation now. And either way, I would put that into a new pull request.