Unifying i387 support

[Jim Blandy]

> > So, correct me if I'm wrong, but here is our current plan, as I
> > understand it:
> [snip]
> 
> Yup, this seems like a good plan.

Okay, we can go ahead with this, with the understanding that we may
need to change the FP register numbering later on.  There is a
possibility that it will make a significant difference in the size of
the debug information if we store FP registers so that the base of the
FP stack, rather than ST(0), is at a fixed position in the register
bytes.  But that's a change we can make later, and it's going to take
a lot of work on the EGCS side before we know which is best.

Anyone want to propose the initial contents of tm-i386reg.h?  Someone
with a newer manual than me, and who knows how 3DNow works?