add benchmark

It would be nice to have a benchmark showing how fast `hrep` and `highlight` are. Ideally it would test `highlight` and `hrep` on two different types of large input:

1. Many input files (thousands? tens of thousands?)
2. A few input files with many lines (tens of thousands? hundreds of thousands?)

This mostly requires just writing a little bit of code to do the benchmarks, as well as code to create (and clean up) the input files before the benchmarks are run.