Choices remain between different negation constructions (simple vs not)

[ajcassell]

To reproduce:

1. Create basic grammar from questionnaire
2. In sententional negation, choose simple, pick "Adverb like particle" and associated questions/orth
3. Then switch to bipartite and fill out this information. You will see that the orth from the simple shows up here. Change the orth to something unique.
4. If you go back to simple, the orth has been overwritten.

I'm not sure exactly where the problem is here, but it is obvious that the values between simple and bipartite (or otherwise) should be mutually exclusive so that in the future, both construction types can co-exist without overwriting. I think it is mostly a UI issue (causing the user to update the orth) but I'm not sure of the downstream effects of them sharing a comp-neg-orth value in the choices file