modeling include_container and transclusions not proving useful

First, as a bug it looks like include_container with no dynamic name set mishandles transclusions.
Considering that include_container is likely to be a transclusion point, this is not ideal.

More generally, include_container has only been used in the h-o layout to glue in each slot.
Transclusions have only been used to glue the virtual h-o layout together with the real one.

Alternatives to include_container that have worked relatively well include:
* just import something from another file
* call dynamic_name around something.

Alternatives for transclusions include:
* inheritance
* doctoring the layout after it is created

Potential things to make include_container useful:
* Support including a list of objects
* Explode the contents of an object and include them. I'm not at all sure that's going to work well, but there are times when I've wanted to include a set of networks etc. I.E. the loop goes on the other side of the boundary from the including container.
* Consider running all the namespace filters and reusing more of the namespace logic
* Provide a way to filter which keys propagate so for example you can use only some of the objects from a h-o layout