I understand that such API will allow to do a lot of things. And it's good and bad in the same time
At the one hand with it someone can implement cool feature like you mentioned before
At the other hand it may bring big inconsistencies into the language, which scares us from design point of view
I've created
KT-55386 just now as a root for some specific discussions about such API, so please write your usecases in it. More usecases we have -> more reasons to consider for our team