Missing optimization: trivial cases

In

  @EStateM.Result.casesOn _x_3
    (fun (a_7 a_8 : _obj) =>
      @Lean.Elab.Term.elabTerm stx t _x_1 _x_1 a a_1 a_2 a_3 a_4 a_5 a_8)
    fun (a a_7 : _obj) => EStateM.Result.error _neutral _neutral _neutral a a_7

the second case could be replaced by fun _ _ => _x_3 (but only after erasure), saving us from generating reset-reuse code. Funnily enough, in the fast path of reset-reuse this is already the produced code since it's clever enough not to re-set any fields when reusing _x_3. Also, cesimp contains an optimization in the other direction, so the new optimization should happen somewhere in between these two phases.