Provide a way to run a cleanup script after a deployment error

CodeDeploy is missing a mechanism for doing cleanup after a deployment error. This could be implemented as another lifecycle event that is triggered only when a deployment fails for some reason. This reason could be e.g. a non-zero exit status of a hook script, a timeout or a user-triggered cancellation of the deployment.

The idea of this originates from [a more specific issue regarding ASG lifecycle scripts](https://github.com/awslabs/aws-codedeploy-samples/issues/42). In that case, an auto-scaling group is left in an unclean state after an error during a deployment.

One remotely similar mechanism for error handling is the `rescue` block in Ansible:
http://docs.ansible.com/ansible/playbooks_blocks.html

Start by reading the CodeDeploy lifecycle event behavior described here and the linked ASG lifecycle scripts issue 42. Compare the proposed failure cases—hook errors, timeouts, and cancellations—with Ansible's rescue-block reference. Done would require an agreed design for a cleanup event and its behavior across those failure cases.

Written by the indexing model from the issue text.