The current behavior of stop-orchestration relies on some assumptions that are not always true:

• There is a server.pid file and it is pointing to a running process
• or, there is a container_id.txt file and it is pointing to a running container
• or, there is a docker container running with the name mongodb-atlas-local

This prevents us from catching the following cases:

• Mongo orchestration is running but we don't have a pid file or it is out of date
• A mongodb or mongodb-atlas-local docker container is running and we don't have a container_id.txt file or it is not named mongodb-atlas-local.
• mongod/mongos were started locally by another process and are running.

We can accommodate all of these cases by using the psutils python library, and shutting down docker containers by image name instead of name.