We need to handle old boot images so that node scaleup works fine for cluster based on older releases < 4.12 .

Update the enhancement with a strawman workable design for handling this.

Tentative proposal:

Change machine-config-daemon-firstboot.service to have special case code that runs the new format oscontainer with `–privileged` and mounts the host root and uses `ostree container pull` into the host repo. This is basically running the new rpm-ostree in a container image. We may run into SELinux issues where we need to run the target in `install_t`.