Why would we (necessarily) need to specify it operationally to begin with?
We can keep deferring this problem to users, but the reality is that they need a solution. Otherwise, bugs end up cropping up in real projects as the paper shows.
Therefore, I argue there has to be a way to guarantee a clear of a memory region in a systems programming language that is not rendered void by some transformation.
I would agree with that if we were not talking about Rust, i.e. how is Rust a "high-level general-purpose language"? It certainly deals with low-level details and systems programming, and even the book, the nomicon and the reference acknowledge this.
Don't get me wrong, I know you are a language team member, but I am genuinely surprised about such a characterization.