Arch-independent double-word clz expansion and a few other tweaks for bitscanning

[Zack Weinberg]

Hans-Peter Nilsson wrote:

> But let's just change [the order of operations in the i386 component of the patch],
 > if nobody (people or ports) disagrees.

I would like to raise a mild objection on the grounds that as far as I 
can tell the code works fine as is.  Without a concrete demonstration of 
a case where it doesn't work, this change seems pointless to me.

Also, before I go and revise the patch, does anyone have any comments on 
the rest of the patch, which needs careful checking rather more than the 
i386 changes do...?

zw