Rather than wait for all containers to come up successfully, immediately schedule a retirement of the old container. We'll still perform a disown'd removal after a successful deploy, but we _also_ schedule a retirement when:
- the new container is run successfully, before we update the container id file
- the new container fails and we run the kill_new command (the schedule is there to ensure removal if the container removal commands fail)
Closes#4576
bf7d6f40ce