Failures after deploy should inform that cloud resources may be used

[johnsca]

If a failure happens during or after sending the bundle to Juju, there may be machines running and cloud resources being used. Depending on the particular failure, the cluster may even be in a functional state. At the very least, conjure-up should point the user at conjure-down to ensure that unused resources are cleaned up.

I'd recommend having the general error handler check conjureup.events.Bootstrapped.is_set() and add a standard addendum to the error indicating that cloud resources may be being used and to use conjure-down to clean up any failed deployments.