Patch for review if someone has a spare moment

["Noel J. Bergman" <no...@devtech.com>]

If folks have a moment, please review the attached patch.  I just want to be
extra paranoid about the change.  It is intended to optimize the two
boundary conditions where a matcher returns ALL or NONE.

The impact of this patch can be seen from:

Without:

31/05/03 23:04:47 Checking Mail1054436556483-30007 with
...matchers.All@1b33a0e
31/05/03 23:06:43 Servicing Mail1054436553472-30006 by ToProcessor Mailet

and

31/05/03 23:51:22 Checking Mail1054439349927-50009 with
...matchers.All@c05a32
31/05/03 23:52:24 Servicing Mail1054439349927-50009 by RemoteDelivery Mailet


With:

01/06/03 02:31:53 Checking Mail1054449100454-1 with ...matchers.All@e51b2c
01/06/03 02:31:53 Servicing Mail1054449100454-1 by ToProcessor Mailet
01/06/03 02:31:59 Checking Mail1054449100454-1 with ...matchers.All@1de4e31
01/06/03 02:31:59 Servicing Mail1054449100454-1 by RemoteDelivery Mailet

The recipient lists were huge, as one might expect from a large, active,
list server.

So far my tests look good, I think the code is right.  As I said, I'm just
being extra cautious.

	--- Noel