Replacing a db_foreach with a tcl foreach loop of a db_list_of_lists because you can run out of db pools if site_node::update_cache is called in the loop for too many parent nodes
Notifications can only be bound to the object_id, so if any contacts instance requests that a group use notifications all instances must use notifications. I've reflected this in the admin pages.
Changing the proc instance_new, now it takes the package-instance-name instead of instance_name, and fall back to current behavior if the package-instance-name is empty