Add strong test coverage of the content model

I am going to unilaterally commit this. Come at me, bro, if you must.

This keeps us honest, ensures our fidelity to the content model as designed and agreed upon, and makes sure we don't screw up minor details during implementation and refactoring, which is very easy to do.

I think it will only help us, not hurt us. If something in the content model changes, tests should break. That proves we are watching.

+1 to RTBC.

Merged into 0.x.

Automatically closed - issue fixed for 2 weeks with no activity.