Currently, probability mass that falls out of the range of TIC histograms does not negatively affect the metrics. @ziz1 suggests having histograms with open edges, i.e. samples falling outside the range will contribute to the first and last bin.
@nec4 any opinions?
