n_features in plot_threshold exploration

[gzentner]

As it stands now, this function colors the dots according to the number of features with a detected TSS at each threshold, regardless of whether it is promoter-proximal. I think it would be more intuitive for the color to represent the number of features with a promoter-proximal TSS (at least, this is how I thought of the plot). Perhaps there can be an option to ask for the total or promoter-proximal feature number, with PP as default.

[rpolicastro]

reopened because referenced incorrect issue with PR #162