Goal is to investigate and implement a way to prepare an ostree commit that would result in unhealthy system.
It's needed in following scenarios:

• After staging new deployment and rebooting, MicroShift fails to make a backup
• After "greenboot reboot" MicroShift manages to make a backup (but it should be made for the previous (rollback) deployment, not current (previous boot)
• MicroShift never manages to make a backup, upon rollback to healthy one, it should create a backup compatible with current deployment (initial, healthy one, from which upgrade started but failed)
• MicroShift upgrades fine, but we add another greenboot script that would just fail causing a rollback

It'd be nice to have one "faulty" commit that'd be controlled in some way to simulate different failures (maybe kernel args? It could also have a logic to fail depending on boot_counter's value).
Maybe instead of dedicated image it could be an "agent" embedded in every commit waiting for specific command.