Increasing the accuracy of Gamma for double float

Dieter Kaiser wrote:
> Hello Ray,
> 
> I would appreciate if we get the more accurate float precicison for gamma. I
> think a lot of more special functions would get more accurate also.
> 
> We had implemented an overflow check for gamma-lanzcos which is now again
> missing. Therefore one test (Problem 196) which check the overflow code failes
> with GCL (not with CLISP).
> 
> Do you would like to implement the overflow check again?

I've added the overflow check to gammafloat and checked in the changes.
Test suite passes as expected.

Ray