intobject.c:int_nonzero() gets called 100 times more than that
in the same test run.  Actually all builtins go through the 
fast slots machinery and not slot_nb_nonzero() so I'm not 
worried about speed (if there is any to be had).

I tried adding a nb_bool to listobject.c (currently it falls
back on tp_as_mapping->mp_length) and I couldn't tell the
difference with 
./python Lib/timeit.py "if [] or [] or [] or [] or [] or [] or [] or [] or []: pass"
So it looks like the places that matter are already fast.