objects_nested_in_text_are_clipped_by_the_type now also nests a 4 x 4 Rstr pixel layer under the TxtA and checks that it is read as an image in the clipping group, next to the shape.