RT benchmark

Took like 16 hours to compile GCC 4.6.1. I found another compiler bug this time in GCC, "static list<listHash*> s_hashes" didn't get initialized before hash classes started calling registerHash(). Hmmm "listHash" is probably the worst name I used it should really be "hashInfo". Anyway I got the benchmark to compile in GCC. Apparently GCC is better at optimizing the reduction function (9.5 - 10 MReduxes/s vs 7.4 - 8 MReduxes/s). Also hashing algorithms got a speed increase except SHA1 and MySQLSHA1, but overall links/second is higher in all cases.