Investigation of signal categories at the fitting level

[VourMa]

This is to keep track of the checks that need to be done to understand the impact of event categorization on the final limits. The checks need to be performed on a boosted and a non-boosted signal mass point - let's take 1000_90 and 280_90 as examples.

• Produce plots for the signal fit and understand how the fit parameters evolve as we go from the most to the least sensitive event categories.
• Calculate the signal yield, the background yield and the background yield in a window of 5σ around the signal peak (reporting how large the window is) for the different signal categories.
• Calculate the limits without removing categories and without the --prune option, and without the -prune option only. The latter configuration requires to drop categories with 0 signal yield. Compare the limits in these cases with the nominal ones as well as with the cut and count limits reported in the optim_results.json file.
• Remove a few categories by hand and keep track of the change in the limits. Compare with nominal results.
• Compute limits on event categories based on different thresholds for adding extra signal categories.