No, this is not planned, and the semantics of this proposal seems fairly confusing to me. We do not always evaluate when branches sequentially, and it’s likely that future versions of Kotlin will use table-driven lookups in more cases than the current implementation, so copying the variable declaration into each branch would be the only way to implement it.
Then the question arises: what to do if a branch does not use the value of one of the variables defined above it? Do we copy the code of the initializer expression into the branch anyway, to ensure that the side effects of that expression still get evaluated? Or do we omit it, in order to reduce code size and improve performance, and lose the predictability? Also, can you refer to a in when conditions, and if you can, when does it actually get evaluated?
In other words, I’d be surprised if this is ever added into the language.