[RFA] convert_doublest_to_floatformat: handle off-range values.

[Joel Brobecker]

> gdb/ChangeLog:
> 
>         * doublest.c (convert_doublest_to_floatformat): If the exponent
>         is too small, treat the value as zero.  If the exponent is too
>         large, treat the value as infinity.

FYI: I checked this patch in. We could do even better (for really
small numbers), but I don't think I will have much time in the near
future and this patch isn't making things more difficult in that
respect. In the meantime, I think this is an improvement over
the previous situation.

-- 
Joel