It does not fit their definition because the whole point of the paper is that a physical system is not computing anything unless an entity is using the system to compute something by encoding the problem into the state of the system and later decoding the answer from the state of the system.
This is a theory, a valid interpretation, highly reliable, scales well, and it commutes. Why isn't it a computation by their definition?
So the question isn't whether soap bubbles can do certain computations, but rather whether soap bubbles can efficiently compute functions that are NP-complete in classical computing models.
That being said, there could very well be models of computing (based on weird physical phenomena) that violate one or both versions of the Church-Turing thesis. Scott has expressed a few times that the possibility intrigues him, though I don't know whether he would bet his life's fortune on one outcome or the other :)
It looks like their definitions are so general that they don't answer their own questions, which by their main motivations (section XI) are fundamentally problems of construction and scale. It's not enough that there exists a theory that proves your physical system computes, you have to know the representation explicitly and be able to scale the system arbitrarily. [Edit:] The problem being that this doesn't show up in their actual definitions.
Maybe I'm reading it too shallowly, though. Can someone who's more well-versed in the background of this paper explain it better? It also doesn't help my confidence that they seem to completely ignore the rest of computer science (mentioning Turing machines only as an aside, and as part of a false assertion about (quantum) Turing machines being the only universal logical systems).