There is no parent context that is passed down to calls in the asset graph and the subsequent calls to wait for the API, destroy bootstrap etc. Sharing the context would make it easier to shutdown the cluster api system and not leak processes.