Delete optimized sigmoid

We have vectorized (and parallelized) portable ops now as of #11912, so we don't need a handwritten optimized op that only does vectorization. (I spot-checked assembly to confirm the portable op is getting vectorized when we build optimized_portable_kernels.)


ghstack-source-id: 7ac8d91
ghstack-comment-id: 3002062946
Pull-Request-resolved: #11939

Author: swolchok

kernels/optimized/cpu/op_sigmoid.cpp

@@ -52,11 +52,6 @@
     - arg_meta: null
       kernel_name: torch::executor::opt_exp_out
 
-- op: sigmoid.out
-  kernels:
-    - arg_meta: null
-      kernel_name: torch::executor::opt_sigmoid_out
-
 - op: gelu.out
   kernels:
     - arg_meta: null

kernels/optimized/optimized.yaml

@@ -307,7 +307,7 @@ def define_common_targets():
     _common_op_test("op_scatter_add_test", ["aten", "portable"])
     _common_op_test("op_select_scatter_test", ["aten", "portable"])
     _common_op_test("op_select_copy_test", ["aten", "portable"])
-    _common_op_test("op_sigmoid_test", ["aten", "portable", "optimized"])
+    _common_op_test("op_sigmoid_test", ["aten", "portable"])
     _common_op_test("op_sign_test", ["aten", "portable"])
     _common_op_test("op_sin_test", ["aten", "portable"])
     _common_op_test("op_sinh_test", ["aten", "portable"])

kernels/test/targets.bzl

@@ -202,12 +202,6 @@ OPTIMIZED_ATEN_OPS = (
         ],
         deps = [":fft_utils"],
     ),
-    op_target(
-        name = "op_sigmoid",
-        deps = [
-            "//executorch/runtime/core/portable_type/c10/c10:aten_headers_for_executorch",
-        ],
-    ),
     op_target(
         name = "op_gelu",
         deps = [