[patch, fortran] Don't initialize components of pointers to derived compounds (PR30793)

[Paul Thomas]

Tobias,

I completely failed to observe that it was a pointer - not only is this 
"crazy", as you put it, but standard defying too.

Also, this now reminds me of where I had seen such behaviour before - 
use of I/O sometimes prevents segfaults associated with an intermediate 
function result.

>
> Ok for the trunk and after a week or so for 4.2 and 4.1?
>
>   

OK

Thanks

Paul