If I understand Neil (KP) correctly, you do in the AI pretty much what you do through the User Interface - define a screen diagramobject on the diagram, with its position coordinates, then separately define each UI Control diagramobject on the diagram, each with its own position coordinates. The UI Control coordinates overlap with the screen coordinates, so that the end result is each UI Control object is sitting over the required portion of the screen.
Did Neil's explanation of the second problem work? Once you have the screen in the Project Browser, you use Ctrl + drag and select the Embedded Objects checkbox to show all the UI elements. Otherwise you get a simple instance or copy of the screen element only.
Does any of this solve the problem? I might not have understood Neil correctly....!