CDFZRAP with last two grids combined

[cschwan]

Commit 1eb397d was supposed to go in a different branch, but I forgot to switch to that branch. However, I didn't overwrite anything so it shouldn't be too bad.

The commit contains the updated root file and the C++ program which combines the last two bins. Performing a convolution with NNPDF31_nlo_as_0118_luxqed before the combination gives

...
bin #26: 3.054914e+00 [pb/GeV] or 3.054914e-01 [pb]
bin #27: 1.433284e+00 [pb/GeV] or 1.433284e-01 [pb]
bin #28: 5.821576e-01 [pb/GeV] or 5.821576e-02 [pb]

and after the combination

...
bin #26: 3.054914e+00 [pb/GeV] or 3.054914e-01 [pb]
bin #27: 1.007721e+00 [pb/GeV] or 1.007721e-01 [pb]
bin #28: -nan [pb/GeV] or -nan [pb]

which are the correct values; bin 26 is unchanged, bin 27 has the average of two combined bins, and bin 28 must be cut away before generating the FK table.

I haven't checked the FK table generation, let me know if there are any problems - there might be because APPLgrid doesn't really combine the grids internally, only when convoluting.

[enocera]

@cschwan Thanks, it works. I've created a separate folder with the fixed data set, that I've renamed CDFZRAP_NEW consistently with the corresponding data implementation, see NNPDF/nnpdf#1068. See also NNPDF/apfelcomb#79

[enocera]

I'm closing this issue as it's been solved long ago.