When attempting to kill a dead processes, do not return failure

[rhatdan]

A race exists where a container engine tells crun to kill a container, process, where the process already died. In this case do not return an error.

Fixes: containers/podman#18452

[flouthoc]

Not sure why is it syncing libocispec in commit, coudl you remove libocispec sync.

[giuseppe]

I am not sure this is correct though, runc returns an error when the container is not running:

$ runc kill foo 9
ERRO[0000] container not running  

I think it should be the caller to ignore the error, or check that the container is still running if kill failed.

[rhatdan]

Ok I think that Podman does ignore it, but an ugly message prints about "open pidfd" from crun, when this condition happens.

[giuseppe]

We can improve the error message when the process is not found but I think we still need to return an error, otherwise there is no way to know if the signal was sent or not

[rhatdan]

Is there a way to silence the error, IE exit with non zero ESRCH but don't write to STDERR?

[giuseppe]

what do you think of this alternative solution: containers/podman#19760 ?