Problem happened when a 1D likelihood shows two intervals below 1-sigma. likelihood_scan.pdf.
This MR allows to draw two lines (or more) in the ExclusionAndBestFit, only one of them with the best fit. exclusion.pdf
An ideal case would be to modify evaluate_likelihood_scan_1d
but it may impact more functions so I restricted myself to modifying the plot_exclusion_and_bestfit_1d
only, for people who are interested in a quick fix.