New versions of fixed and floating point packages

I just posted new versions of the fixed and floating point packages for
VHDL.

http://www.eda.org/fphdl/vhdl.html

In this version the "read" (and oread and hread) procedures can accept
an "_". Like this:

variable L : line;
variable uf : ufixed (7 downto -;

L := new string'("0000_0000.0000_0000");
read (L, uf);

In the floating point packages a bug was found when multiplying normal
numbers by denormal numbers.

David Bishop