This replaces the lazy_bdecode decoder with the more efficient decoder from libtorrent. More details on performance here: http://blog.libtorrent.org/2015/03/bdecode-parsers/
Only basic testing has been done, by poking it with the test_dht.py script. However, that is where most of the bdecoding happens, so it covers a significant portion of the updated code.
use the more efficient bdecoder from libtorrent
rebased on top of master