I am having difficulty modelling an assembly that has a top-down modeling approach, but where the parts need to move after being modeled in-context.

For example, I have attached a model of a simple metal box with a hinged lid.  The lid and hinge are in their own subassembly.  The size of the lid is related to the size of the body so it fits over nicely, and the hinge is mated to the top of the lid, and also to a plane that represents the rear of the box.  I have also modeled a hole in the front of the box and the front of the lid, where the lid hole has been made using "convert entities" from the body hole. 

Now I want to be able to make the lid assembly a "flexible assembly" to allow the hinge and lid to move, but I need all of the existing features (such as the hole in the front of the lid) to still be in their correct locations.

I have struggled with this for a while on different projects, and tried different solutions, including adding another configuration in which the original instance of the lid assembly is suppressed and I add in a second instance of the lid assembly.  But whatever I do something always seems to go wrong: Either I have mate errors (overdefined), or the lid subassembly becomes rigid (even though it should be flexible) or something else.  Sometimes everything works great until I add a new component to the subassembly and then things go wrong for reasons I can't explain.

How do you handle this sort of situation?