Several updates and new features

This MR includes several changes, fixes, improvements and new features.

• Reworking how input variables are treated in the code. Instead of passing a list of variables in the config file and the corresponding binning in Histograms.py the variables are now defined directly in the config file via VARIABLE blocks:

  VARIABLE                                                                                                                                                                                                           
        Name = Variable_1
        Label = "Var 1"                                                               
        Binning = 25,-4,9

• TRExFitter's TransfoD style binning optimisation is now available via the config option BinningOptimisation. It can be used in the MODEL block.
• The script to generate ntuples for the CI scripts has been extended to also include a "truth" tree.
• Fixing indentation in AtlasLabel.py. Was 6 before, is 4 now
• Improving code in several files based on pylint suggestions
• Adding an implementation of a swish activation function in CustomFunctions.py
• Adding a file for custom layers, CustomLayers.py to be used later for extension purposes
• Moving several functions into HelperFunctions.py
• Fixing issue with MVA-Injector where exceptions were not piped through. Now the injector should stop if one of the child processes encounters an error -> most likely indicating a problem with the code itself. This might need some tweaking in the future
• Completely getting rid of Histograms.py. This file no longer serves a purpose because the creation of variables is now handled centrally in the config file.
• Cleaning up some of the plotting scripts and moving functions around
• Fixing some typos in root scripts
• Restructuring the CI pipeline. Binary classification and multi-class classification are now run as separate "blocks".