Lean proofs for the Action circuit soundness sound useful. Is this mostly a machine-checked restatement of the existing spec, or does it already catch any mismatches?