Allow OCI's dynamic inventory plugin to scrape stopped compute instances

[smckend]

I'd like to propose a change that allows for hosts that are not in a "running" state to also be added to Ansible's inventory. Doing so would allow for more granular filtering and, more specifically to my use case, it would permit starting/stopping instances before and after Ansible plays that would otherwise need to be manually started.