More backreference performance improvements [take 2]

[Isamu Hasegawa]

From: "Bonzini" <bonzini@gnu.org>

> count).  My 60% optimization does nothing more than limiting the lower and
> upper bound of the inner loops in sift_states_bkref.

I meant I'll also fix the loops you gave the optimization.
I'm afraid we will need to do profiling again if the algorithm is changed.
Of couse, when your fix will have still effect for the revised algorithm,
let us use your fix.

Thanks,
-- 
Isamu Hasegawa
IBM Japan, Ltd.