This appears to be fixed as of April (at least for Apache). [0].
[0] - https://github.com/nginx/nginx/commit/365694160a85229a7cb006...
I was about to say, the bug here isn't in the protocol, it's that memory use isn't being counted & limited as it should... and, yeah.
I'm a bit surprised this happened to Apache, though. APR uses pool allocators. That should be easy enough to track and limit...
This is already something that is known, and if you're able to be targeted by this (which is not the majority of users) configure your httpd differently.