Changes to hashing that solve the xinetd boolean benchmark in 14s (from ~25min)....

Changes to hashing that solve the xinetd boolean benchmark in 14s (from ~25min). Switched to standard hash_set, hash_map, new hash for the vector of node values (from boost), changed the hash for nodes to be over id's, all the hash values are now size_t. The parser is down from 11s to 10s on the benchmark, so most of the solve time is parsing and we need to figure this out.