Automatic metric comparison

[travisdriver]

I'm not sure how hard this would be, but thought I would just see if it's possible.

It would be nice if reference metrics could somehow be stored on master, and then these could be compared with the metrics output from each pull request (just computing the difference between all the values would probably be sufficient). On merge, the reference metrics on master would be updated.

Could be a nice feature if possible, but either way the automatic benchmarking is still an amazing tool :)