Sharing aggregate states between different aggregate functions

Simon and I have been going over some ideas about how to make improvements to aggregate performance by cutting down on the duplicate work that's done when 2 aggregate functions are used where one knows how to satisfy all the requirements of the other.

To cut a long story short, all our ideas will require some additions or modifications to CREATE AGGREGATE and also pg_dump support.

Tom came up with a more simple idea, that gets us some of the way, without all that pg_dump stuff.

This basically allows an aggregate's state to be shared between other aggregate functions when both aggregate's transition functions (and a few other things) match

There's quite a number of aggregates in our standard set which will benefit from this optimisation.

Please find attached a patch which implements this idea.

The performance improvements are as follows:

Regards

David Rowley