It seems that this paper addresses the same limitation I once tried to describe in thread Variable dependent types? In particular this quote
The key idea in this work is to encapsulate a reference to an object that is not (yet) a borrow
reminds me of some ideas in this post.
Definitely I agree that this is an issue Rust should tackle and that the ideas exposed in this paper are certainly worth considering. (The normal procedure should have been to write an RFC but I'm a afraid that now it is probably a dead-end).