Rework ROOT file reading

[AndreasAlbert]

Relates to #99

The current purpose of ROOTFileReader is to provide a tool to extract arrays of numbers from histogram and graph axis, or possibly trees. Most of these features are implemented in uproot in a much nicer fashion, so there is no point for us to maintain effectively duplicate code. Instead of using ROOTFileReader as a ROOT-to-array converter, we should use it as a ROOT-to-hepdata_lib.Variable converter.

The following operations can definitely be handled by uproot:

• TGraph -> array conversion.
• TH1 -> array conversion.
• TH2 -> array conversion. Caveat: need 1D array format for dependent variable consistent with what hepdata expects.
• TTree -> array conversion.

The only operation we need ROOT for would be to retrieve histograms from TCanvas primitives.