run functorch decomps after functionalization when enabled (#83992)
This is a short-to-midterm fix for https://github.com/pytorch/pytorch/issues/83923. By running functionalization before decomps, we guarantee that functionalization won't have to see any primtorch view/inplace ops like `broadcast_in_dim`.
This will only really be a problem if there's a function in the decomposition table that decomposes a functional op into mutations. If that comes up later, we'll need to revisit https://github.com/pytorch/pytorch/issues/83923.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/83992
Approved by: https://github.com/ezyang