latest RPM

[Matthew Fluet]

> Matthew, when I shipped the latest RPMS, I forgot to mention the changes to your 
> stuff.  There was only one: I added stuff to x86-translate for Word{8,32}_neg.
> Another thing that I changed in backend.fun that you might be able to take
> advantage of is that void operands will never appear in machine-output.

I saw the Word{8,32}_neg.  

The only place where I make use of Void is in MLton_eq, where two voids
are always equal.  If you're confident enough about never propagating back
voids to the codegens, we could change the MachineOutput.Operand.t type to
reflect that.