Deterministic hash table benchmarks

[Jason Orendorff]

I wrote a benchmark and tested an implementation of Tyler Close's
deterministic hash table against some conventional, non-deterministic
ones.

Tyler Close's table is faster, but it uses more memory.

Details and graphs:
  https://wiki.mozilla.org/User:Jorend/Deterministic_hash_tables

Source code:
  https://github.com/jorendorff/dht

These results are very raw. Take a look at the code. See if there are
simple ways to improve the performance of either hash table. Or
explain to me why CloseTable is faster even though it's doing more
work. Or why the DeleteTest results are so weird.

-j