Performance / benchmarking

[timkurvers]

As discussed with @fallenoak in #76:

Maybe we could get something set up in the test suites for Blizzardry to try benchmark load performance for a specific, complicated M2 (like Illidan)? For example: load Illidan from scratch 10 times in a row, average the load time, report it back on the screen. Then we'd at least have an easier way to compare with previous results on the same systems when trying to optimize code paths running during parsing.

The main thing to tackle here would be the benchmarking parameters, how to opt in for these types of tests and where the fixtures (for example Illidan.m2) reside.