Re: range_agg with multirange inputs

This patch has been committed.  I split it into a few pieces.

On 12.03.22 04:18, Paul Jungwirth wrote:
> On 3/10/22 14:07, Chapman Flack wrote:
>> When I apply this patch, I get a func.sgml with two entries for
>> range_intersect_agg(anymultirange).
> 
> Arg, fixed.
> 
>> In range_agg_transfn, you've changed the message in the "must be called
>> with a range or multirange"; that seems like another good candidate to
>> be an elog.
> 
> Agreed. Updated here.

I kept those messages as "range" or "multirange" separately, instead of 
"range or multirange".  This way, we don't have to update all the 
messages of this kind when a new function is added.  Since these are 
only internal messages anyway, I opted for higher maintainability.