Union type as ADT argument about alpaca

[j14159]

I wonder if this is similar to #34 - wouldn't surprise me if we could simplify the handling of ADTs and their unification while solving both, it's a bit heavy-handed at the moment. I'm going to try to find some time to sort out #25 over the next week (I suspect it's just a case of furthering generalization) and then tackle this (unless it's already solved by then ;) )

[danabr]

I have a prototypical fix here: master...danabr:fix/complex-types

I say "prototypical", because the code needs cleanup, and I'm not sure I should do type lookup like this. It also happens to fix the example in #34, but there are more complex ways to use aliases than that example which I'm sure will break even with this simple fix.

[danabr]

I've updated the complex-types branch with something that works even for type aliases and type unions.

However, there is something weird going on. unify_adt expects two cells as its first arguments, but there are occasions when it gets something that is not a cell. You can see this by removing the "is_pid(C2) test and running the tests.

[j14159]

Closing due to PR #49