Again, I'm not a physicist, but I think it's telling that the validity of continuum mechanics [0] depends on a model and multiple assumptions. I have no trouble agreeing with your original statement, "with enough marbles, the discrete distribution behaves like a continuous distribution," when speaking mathematically, but mathematics by its nature is an idealized model of the world - I'm not willing to accept that it's objectively representative of physical reality. The discussion we're having here is one of metaphysics, so it feels a bit like the height of hubris to use mathematics as the tool for describing objective reality, because metaphysically, we cannot say that mathematics is anything other than a tool we've constructed for approximating the model of the world as we understand it. Considering the subject of this post is about retrocausality, we're already throwing out some pretty wild ideas, so I think it's a bit hubristic to dismiss them by citing a branch of mathematics that assumes the existence of countably infinite sets [1].
[0] https://en.wikipedia.org/wiki/Continuum_mechanics#Validity
[1] https://en.wikipedia.org/wiki/Constructivism_(philosophy_of_...