Draft "Unsafe fp optimizations" project description.

[Gabriel Dos_Reis]

| > In addition, in many architectures, and in particular on the x86, you
| > are using extra precision and range for intermediate computations, and
| > in this case, the number of overflow cases is drastically reduced.
| 
| Good point - I didn't consider that possibility.

If that point were to be considered or included in the discussion, it
would be a good thing to provide a list of supported plateforms where
intermediate results use or do not use extra precision (e.g. x86-like
vs. Sparc-like).

-- Gaby