Contravariant instances for `Op`

I noticed there are a number of possible contravariant instances for Op if we make some Arrow assumptions but currently none are there. Do we feel like any of these would be beneficial? For example, something like...

private[data] trait OpDecideable[Arr[_, _], R] extends Decideable[Op[Arr, R, ?]] with OpContravariantMonoidal[Arr, R] {
  implicit def Arr: ArrowChoice[Arr]
  implicit def M: Monoid[R]


  def sum[A, B](fa: Op[Arr, R, A], fb: Op[Arr, R, B]): Op[Arr, R, Either[A, B]] =
    Op(Arr.choice(fa.run, fb.run))
}

private[data] trait OpContravariantMonoidal[Arr[_, _], R] extends OpContravariant[Arr, R] with ContravariantMonoidal[Op[Arr, R, ?]] {
  implicit def Arr: Arrow[Arr]
  implicit def M: Monoid[R]

  def unit: Op[Arr, R, Unit] =
    Op(Arr.lift(Function.const(M.empty)))

  def product[A, B](fa: Op[Arr, R, A], fb: Op[Arr, R, B]): Op[Arr, R, (A, B)] =
    Op(Arr.compose(
      Arr.lift((M.combine _).tupled),
      Arr.split(fa.run, fb.run)))
}

private[data] trait OpContravariant[Arr[_, _], R] extends Contravariant[Op[Arr, R, ?]] {
  implicit def Arr: Arrow[Arr]

  def contramap[A, B](fa: Op[Arr, R, A])(f: B => A): Op[Arr, R, B] =
    Op(Arr.compose(fa.run, Arr.lift(f)))
}

Start by reviewing Op and the Arrow, Contravariant, ContravariantMonoidal, Decideable, and Monoid abstractions referenced in the proposal. Determine whether the proposed instances belong in Cats and what assumptions they require; done means a maintainer-approved scope, implementation location, and tests for the selected instances.