Add CDFZRAP dataset

[cschwan]

This branch add the CDFZRAP dataset from arXiv:0908.3914. The data can be found at http://hepdata.cedar.ac.uk/view/ins856131 and https://www.hepdata.net/record/ins856131.

Issues:

• static scale choice - how was the QCD APPLgrid generated?
• photon recombination radius not known - Rivet analysis sets it to 0.2
• EW data unavailable (?)
• transverse momentum cuts are not complete, they depend on the rapidity range according to the Rivet analysis there are no cuts on the leptons
• mg5_aMC@NLO does not generate a HwU fixed in revision 1011
• PineAPPL needs support for antiprotons fixed in master

[cschwan]

Here's the comparison against the results from the NLO QCD APPLgrid (from the file CDF_ZRAP_MCgrid_500M.root), convoluted with NNPDF31_nlo_as_0118_luxqed, normalised to the APPLgrid results (negative values indicating that the APPLgrid results is larger). The numbers shown are differences in per mille:

8.535916  10.005214  14.431520   8.646862  12.950925   7.502359
6.399847  11.077829   5.718108  13.185019   5.856358   6.904448
2.322006   6.258221  10.249563  -1.218714   6.203596   8.293686
8.344370   5.544752  -0.262706  10.459825   8.166978  -0.173896
7.329547  16.667900 -17.506385  28.797405

In this case I don't think we can claim to have agreement. I wonder what the source of the differences is.