Just out of curiosity since it irked me when I was writing tests for normalization. Why do you use done() in synchronous tests?

@darsain we use the --async-only flag with mocha. This helps to catch subtle bugs, like not passing an error callback to an actually-async test. It is a hold-over (and I'd say a good one) from my dislikes with mocha and my time with lab. I really think using function arity to do async/sync magic is bad practice.

Looks like everyone is too busy to review this. Going to merge.