I'd have to try, but have never encountered something like it on btrfs (though to be fair I've had many other issues and bugs with it over the years!)

I understand the logic, but typically I've seen filesystem implementations block writes once metadata volumes become close enough to full. Also, and I don't know if this is a thing on modern filesystems, you used to be able to reserve free space for root user only, precisely for recovering from issues like this in the past.