Real.maxFinite bug

[Stephen Weeks]

> Yes.  I'm sure that the Real.isFinite comes from the C routine.  Didn't we
> talk about this long ago?  I don't see any thing in the saved mails, but
> I thought we had.

We did talk about how to use Real.isFinite to compute Real.maxFinite back in
June 1999, which led to the code that is now in basis-library/real/real.sml.
But I don't see anything about how to define Real.isFinite.  Anyways, I've now
checked in the change, which fixed the bug.  I still need to look into whether
the "reify" function in real.sml is doing anything useful.