Use airspeed velocity for benchmarking

[cpcloud]

Since vbench was (at least in part) an inspiration for Airspeed Velocity (asv), I think getting something like this set up for pandas would be useful and extremely helpful for us and for asv. We have a ton of vbench code and we'd probably be a great project for pushing the boundaries of and finding edge cases in asv. This would also be a great for someone to get to know the scale of datasets for which pandas was designed.

Here's a nice demo of astropy using asv: http://mdboom.github.io/astropy-benchmark