As of we will need to setup two things for each AWS Elasticsearch endpoint:
A lockable resource with the appropriate label
An environment variable with the appropriate name, and the endpoint URL as its value
We could just use the lockable resource name as URL and avoid configuring each endpoint twice. As a bonus, this would allow to configure multiple resources of each type (multiple ES 5.3 clusters for example). Not really useful to us, but a nice bonus nevertheless.