Add a new method to pre-selection and final-selection. Also add some features to do_plots

[tomfournier]

Hi,
I have written a new method to run the pre-selection. This method has the same properties of RDFanalysis but takes the process_name as a second argument in analysers and return the dataframe and a list of histogram or TParameter that will be written in the output file in a TDirectory called custom_objects.

In custom_object, the objects to write will be put in sub-TDirectory depending on their class (TH1D, TParameter, etc.) to be more readable.

I called this new class RDFgraph because I wanted to take the advantages of RDFanalysis and build_graph and I named the TDirectory custom_objects but if you have more inspiration, you can change the names.

During the final-selection, a dictionary similar to histoList can optionally be added to include the histograms in the final-selection output file called customHist This new variable takes this form:

customHists = {
    'leps_iso': {'name':'ConeIsolation', 'title':'I_{rel}', 'xmin':0, 'xmax':10},
    'leps_no':  {'name':'n_leptons', 'title':'N_{leptons}'}
}

to modify the histogram. If you don't want to modify your histogram, you can just leave the dictionary empty.

I made this method to be able to easily make a cuflow and histograms before a filter since RDFanalysis only return the branches after all the filters.

For the final-selection, we could use an eventual cutflow that is in custom_objects/TH1D to add new cuts from cutList but I think it's too difficult for me to implement it.

I also modified do_plots to have a strict x-range to not display bins with zero content and to set grid on the plot.
To do this you can add strictRange = True and setGrid = True to the plot script.
This apply to all the variables but we could make a list like for rebin to do implement it on chosen variables.

We could also make a dictionary with rebin, grid, strictRange and other parameters for all the variables in this form:

variables = {
    'zll_m': {'rebin': 2, 'strict': False, 'grid': True},
    'zll_p': {'strict': True, 'grid': False}
}

with the keys optional.

I ran it on my setup and it works fine but I don't know if it works for other setup or if I followed the coding conventions. I timed RDFanalysis and my new method and they take the same time to run so I don't think there is much optimization to do.

If I was not clear enough or you need more information, I am available to answer your questions. I hope my modifications can be implemented as I think they could be useful for FCCAnalyses users.