CHANGES IN FIXFLOAT LIBRARY 1.1 -> 1.2

1 - Bugfix in ffcos routine.

2 - Better performance of signed multiply by IMUL & IDIV 
    instructions instead of sign analysis. 

3 - Inline assembly versions of ffmul, ffsmul, ffsdiv, ffdiv have been 
    added to remove CALL / RET instructions when multiplying/dividing 
    from C/C++.

    The following Assembler macros have been added to make provide the
    same effect in asm files :

    MFFMUL, MFFSMUL, MFFDIV, MFFSDIV

4 - The divide functions ( not macros or inline asm functions ) have been
    made foolproof so that they cannot overflow 32 bits and generate 
    exceptions. Instead largest possible positive or negative number is
    returned.

5 - A function to produce the argument of a vector is provided.

6 - A function to solve 2:nd degree polynomials is added. Works with 
    64 bit intermediate results.

7 - A function to approximate 'sqrt( a*a + b*b )' which executes in approx
    18 clocks has been added.

8 - Documentation have been extended.

9 - An extensive C example demonstrating the functioning of approx 80% 
    of the provided functions is included.

    
  
