Dear LC,
That is a very subtle point you make there and I'm not sure I am skilled enough in this domain - I am probably not :) - to be able to bring the discussion further. But at least so as to clarify what I have in mind, I will bring the following example, which I realize is built on a different framework than the one you specified. There exists a mathematical proof for why and how phase transitions happen , but it relies on a lattice treatment (Sinai and Berezin). Until today, there is no treatment that can show the same thing for systems with more degrees of freedom. The emergence here is given by the extra idea that needs to exist in order to complete the proof. This extra idea is what adds insight and moves the pieces of the demonstration together in the right order, so they compose something that was not necessarily and sufficiently implied by the existing theory. That being said, i fully agree that the underlying natural models expressed by mathematics show emergent properties.
Cheers,
Alma