[gfortran] PR15206 RRSPACING wrong for zero

[Paul Brook]

> > On a related issue, the existing rrspacing code to handle denormals is
> > broken. This occurs when x < TINY(x), ie. expn == 0, frac != 0
> > I suspect the right shift should be a logical (unsigned) shift, not an
> > arithmetic (signed) one.
>
> Should I close the old bug and add a new one where we keep track of this
> or should I change the summary line and add this information?

A new one please.

Paul