self-bench vs ml-agents

Compare self-bench and ml-agents using verifiable public signals, licensing and project quality.