tinystat is used to compare two or more sets of measurements (e.g., runs of a multiple runs of
benchmarks of two possible implementations) and determine if they are statistically different.
It's inspired largely by FreeBSD's ministat (written by Poul-Henning Kamp).