related_post_gen benchmark

[PMunch]

Compiling with LTO and removing a couple superfluous things brought it down to 
sub 50ms for me. But I think my machine was possibly quite a bit faster than 
theirs because it ran the original in about 75ms (or they messed up during 
compilation). It seems to be mostly to measure the performance of Table. But 
there are certainly things which can be optimised there.