Create time tests / benchmarks for Filelist

[Cedric-Boucher]

I might not do this. The speed of certain methods of Filelist are fully compute / implementation dependent, and those could definitely be tested for speed, but others are very much I/O dependent and probably can't easily be improved. This could be useful for determining good constants for multithreading group sizes, or testing whether multithreaded methods are actually faster than their singlethreaded counterparts, on the same hardware.