benchmarking Poly/ML & floating point

[Stephen Weeks]

> It can screw you by making the code inside of Real_equal being
> 	test if x1 != x2
> 	if true jump
> 	load %eax with 1
> 	return
> label:	load %eax with 0
> 	return
> 
> I.e., you can be screwed in either place.

True enough (although it doesn't appear to happen, and we can actually check,
since that code is only compiled once).  Anyways, I guess the best thing is to
stick with C #defines and to note that we aren't compliant in the C backend.
Since we're moving to the x86 backend, this may be reasonable.