Investigate Zincati's compatibility with systemd inhibitor locks

[kelvinfan001]

From #245 (comment):
Zincati might want to proactively monitor for systemd inhibitor locks to ensure that shutdown blocks are handled smoothly and behave as expected; in particular, we should ensure that the rpm_ostree actor and the rpm-ostree daemon do not block and that they at least notify the update_agent actor that an error occurred (If rpm-ostree daemon does not block on the reboot, would we have an unlocked staged deployment after a failed reboot attempt?).
Additionally, I've observed that even inhibitors locks owned by root do not prevent Zincati (and rpm-ostree) from finalizing the reboot.
Depending on findings from the above, handle inhibitors locks properly and/or add documentation for how to use systemd-inhibit to prevent Zincati from finalizing.

[kelvinfan001]

With coreos/rpm-ostree#2862, rpm-ostree finalize-deployment will respect systemd inhibitors in block mode, and immediately return an error upon detecting blocking locks. Care has been taken to keep the staged deployment "locked" even after a failed reboot attempt.
For now, Zincati will be blocked from rebooting the machine for as long as a block inhibitor is present, as rpm-ostree does not yet have a mechanism to override them.