What is "evaluation" vs "sample" vs "call"?

[ryantate]

The output of bench begins like so:

Evaluation count : 120 in 60 samples of 2 calls.

In this context, what is an "evaluation," what is a "sample," and what is a "call"?

I could not find this in the docs. It would be nice if it could be (for example) in the README. Or if you want to just explain it in a comment here I'd be happy to submit a PR.

(Thanks for any guidance. Perhaps this obvious to people who are more seasoned than I am at benchmarking, but trying to Google around for answers I could not find any standard definition for these terms.)

(PS Thanks so much for Criterum, it's highly useful!!)