Send (OOM) kills to Sentry

[MrSerth]

Currently, when Poseidon is killed, e.g., by the OOM killer, we do not get notified nor any Sentry event. Ideally, we would add some hook or other mechanism to get informed whenever Poseidon quits unexpectedly.

[MrSerth]

In order to address this issue, we want to tackle two different aspects:

• First, we want to add an ExecStopPost, checking the exit code of Poseidon and potentially creating a Sentry issue.
• Second, we want to formalize our memory monitoring introduced with Introduce debug memory monitoring #457, so that the threshold is customizable and can remain in the source code. This should allow enabling / disabling the monitoring at all. If a memory leak is detected, the information is written as an error to the log and sent to Sentry (if enabled).

[mpass99]

Should we also address setting up the systemd watchdog within this issue (or create a separate issue for it)?

[MrSerth]

Let's extract it to a dedicated issue, since it is not directly related to the OOM kills tracked in this one.