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).