Add new derived curve - frequency of superior performance

[JohnHalleyGotway]

Enhance METviewer to add a new derived curve type for the frequency of superior performance (FSP).  The challenge here is letting the user specify how close statistics can be for them to be considered a tie.  In the plot_tcmpr.R script, that tolerance varies by forecast hour, but that probably isn't realistic via METviewer.
Also need to keep track of what is "better" for each statistics by defining what the "perfect" score is... usually 0 or 1.

Recommend that we also enhance the scorecard methodology to color the cells by FSP values too.
Refer to plot_tcmpr.R script:https://github.com/NCAR/MET/tree/develop/met/scripts/Rscripts

This issue was discussed during a meeting with METplus staff (John HG, Tara, Barb Brown, Eric Gilleland, etc) and NOAA/EMC staff (Ivanka Stajner, Jason Levit, and Fanglin Yang) at NCWCP on 02/19/2020.