Container orchestrator: better handling of machine suspension and hibernation

There is anecdotal evidence that running Containers will not survive prolonged machine suspension, or hibernation. I have seen cases where Containers ended in "Unknown" state after resume.

Unknown state is final in the current implementation. We should consider if we can relax this limitation and handle hibernation better, e.g. by enabling low-frequency probing for containers in Unknown state and allow their transition back to Running and other relevant states.

The issue names no files, tests, or entry points. Start by locating the container state handling and its treatment of Unknown after resume; done requires an agreed behavior for suspended or hibernated containers, including whether probing can return them to Running or other states.

Written by the indexing model from the issue text.