Performance tracing

[ayazhafiz]

I am thinking of adding performance/efficiency tracing to the library.

Basically the idea is to have a set of scripts that run on every CI build and
log data about how long a set of executions took and how much memory was
consumed.

The important thing here is to make sure that the input we are checking is
future-proof and will have a large variance on system resources. So I am
thinking we run over a subset of test files, in determinstic order, in serial.

It might be too early to add this for it to be very useful. But that's okay; its
better to have it early and it not be useful than to add it later and miss a lot
of things that may have been useful to us.

After this and some other issues are done, 0.2 should be released. 0.3 will
focus on stabilizing evaluation and adding more eval features.