WASM PVM: `store_get_nth_key` is probably unsafe

The list function exposed by Irmin does not depend solely on the shape of the tree, but also depends on the context. More specifically, there is an interference with the cache of the context, meaning the result of the function won’t be the same if there is a commit just before the call to the host function…

You guessed it, we have at least a hash inconsistency to deal with.

We should hard deprecate the function, not use it, and potentially turn it into a no-op in the future?