Create tests to measure performance

[troyhunt]

This is related to #41 but I thought I'd break it out into its own issue as it's a discrete unit of work. Without tests, it's very hard to tell if even a really minor tweak to a regex (or similar) is detrimental to performance. I'm not sure of the best way to do this, but we need even a really rough order of magnitude test suite that can pick up on this sort of thing. The change referenced in the aforementioned issue has hit perf by up to 1000x (33 emails per sec in the 300s report versus 3,091 before the change), so I imagine it won't be hard to detect something that varies perf by that much. Hopefully we can get a more finely tuned measurement model together and even start chipping away further at the best perf stats we've achieved.

[jaimevisser]

It would be interesting to have some lifelike dataset to run tests against.

[troyhunt]

I actually generated a set of sample data back in #15 which I've now added to the readme. For brevity, it's here: https://mega.nz/file/Ls8U1ADK#c1We1C_CZi44P0k3OB8YpNVN7HMM3gE_4-fH06E454c