For testing things like
SLIDER-82, it'd be good to have a history where we can list events for each component (e.g. requested containers, released containers, failed container, escalation). This would be a structured log which could be viewed in the REST API for each component, and parseable for the test runner to get a view of what happened to get the cluster to its current state;
Could be good as an HTML view too, so you can see what's going on. We do end up looking in the AM logs for this kind of information, after all.