[gfortran,patch] Fix intrinsics used as actual arguments

[Paul Thomas]

FX,

>> If it is a trivial backport (15 minutes and regression test),
>> then I would think that it is worth. If it requires any effort
>> then I think the port is not worth it.
>
>
> It turned out to be more than that, as I originally had to make a 
> succession of commits that are not backporting trivially (and also, 
> the big diff to intrinsic.c is hard to backport automatically). So, I 
> didn't backport it.

With branching so imminent it seems utterly pointless to fret about it. 
I stopped backporting a couple of weeks ago. After all, if 4.1 is 
identical to 4.2, what is the point of releases?

Lazily

Paul