due to singleton usage we may be reading wrong lastEventId, but it is
not really needed so let's siplify the logic and use the existing
AnsibleReturnValue for keeping track of it.

Instead of parsing the current event we can reuse the same logic as used
for async playbooks, check the "status" artifact to see if ansible is
still running or not, and check if all events have been processed

we rely on this exact behavior of ansible-runner, let's document it