constant switch test bug in x86Translate

[Stephen Weeks]

> Sorry about that.  Here's the full directory.

More progress, but not there yet.  Here's the new error.

   x86 code gen starting
      outputC starting
      outputC finished in 1.720
      outputAssembly starting
	 translateChunk raised
      outputAssembly raised in 354.570
   x86 code gen raised in 551.120
Compile SML raised in 1060.450
mlton: toX86Operand: strange Offset

I put a snapshot at http://www.star-lab.com/sweeks/src.tgz

If you can get it to a point where a G0 self-compile succeeds, that would be
great.  I expect the resulting G1 to fail, because of the aforementioned nasty
bug.  I have alot of the CPS simplification passes commented out in an attempt
to localize the bug.