1. Why isn't observability setup to monitor the queue depth and alert?

2. Why aren't the CI runners self-healing?