Assigning priorities to GNATS reports

[Joe Buck]

> Gerald asked me to write down a policy for gnatswrite.html how to
> mark regressions. Unfortunately, there still seems to be some confusion
> regarding bugs that are present in the 3.2-branch and in mainline.
> Shall we mark them as
> 
>   [3.2/3.3 regression]
> 
> or just as
> 
>   [3.2 regression] 

We need to distinguish those bugs that are fixed in the mainline but not
in 3.2, from those that are present in both.  So I like it the way that
it is now.