Currently the expiry mechanism implemented in the Amazon EC2 blueprints is much closer to a recycling mechanism; resources aren't guaranteed to close after the given time has passed, because they can only be closed when Drydock is executing logic on them.

Instead, we should provide some mechanism for Drydock resources to be checked periodically (most likely through Chronicle). This would allow us to optimize the use of EC2 instances (so they shut down after 55 minutes, rather than spinning up and shutting down lots of instances over an hour because of builds completing quickly).