flatpak or firejail would have protected you from this vulnerability, not sure what they're on about here. They are 100% proof against everything of course.
I couldn't reproduce the tty example, but it might as well be a mistake on my side. Other than this, the sandboxing spec itself is as safe as I'd expect. I reckon that Wayland applications not packaged to require $HOME access or some dbus services are not known to escape the sandbox. This seems to be the case of Firefox, afaict.
