If one looks at the demo time lapses in the Wiki, on notices that there is a clear horizontal barrier between the area where the print head moves and the printed object itself. The print head and its nozzle barely covers the printed object.
So instead of printing a purge tower or moving the head into a parking position, the app could instead remove the print head using some smart image processing algorithms (the upper section of the image doesn’t change light/shadow wise when the head isn’t „there“; yes one would would see the head moving in the reflection on the bed, but one could live with that … or image process that as well.).