Well you don't need to have nonlinear evolution to get what alpineidyll3 is saying. It's sufficient for the observable algebra of macroobservables to be commutative. This allows the evolution to be linear and have no interferences.

The QM is "an island in theoryspace" idea isn't strictly true either. QM is one among an entire family of probability theories. It's only rigid when considered purely from the point of view of Probability theories based around vectors in Hilbert space. However considered as part of OPTs in general there's nothing that makes it difficult to modify.