Update yaml syntax to use kernel instead of dispatch

Summary:
X-link: pytorch/pytorch#104070

Based on this [code search](https://fburl.com/code/gjcnw8ly) (*.yaml with `dispatch: CPU:`), update all files found to use

```
kernels:
    - arg_meta: null
      kernel_name:
```
instead of
```
dispatch:
    CPU:
```
 ---
## Code changes:

- `fbcode/executorch/codegen/tools/gen_oplist.py`
  - Strip ET specific fields prior to calling parse_native_yaml_struct
 ---
## Files edited that are not `*functions.yaml` or `custom_ops.yaml`

- fbcode/executorch/kernels/optimized/optimized.yaml
- fbcode/executorch/kernels/quantized/quantized.yaml
- fbcode/executorch/kernels/test/custom_kernel_example/my_functions.yaml

 ---
## Found Files that were not edited

**Dispatched to more than just CPU**
- fbcode/caffe2/aten/src/ATen/native/native_functions.yaml
- xplat/caffe2/aten/src/ATen/native/native_functions.yaml
- xros/third-party/caffe2/caffe2/aten/src/ATen/native/native_functions.yaml

**Grouped ops.yaml path**
- fbcode/on_device_ai/Assistant/Jarvis/min_runtime/operators/ops.yaml

 ---
**Design Doc:** https://docs.google.com/document/d/1gq4Wz2R6verKJ2EFseLyPdAF0wqomnCrVDDJpRkYsRw/edit?kh_source=GDOCS#heading=h.8raqyft9y50
ghstack-source-id: 193204305

Reviewed By: larryliu0820

Differential Revision: D46952067

fbshipit-source-id: 79528bb751faa489fbde99a01c1950281e94045d

Author: Jack-Khuu

codegen/tools/gen_oplist.py

@@ -5,6 +5,7 @@
 from typing import Any, Dict, List, Optional, Set
 
 import yaml
+from torchgen.executorch.parse import strip_et_fields
 
 from torchgen.gen import LineLoader, parse_native_yaml_struct
 from torchgen.selective_build.operator import SelectiveBuildOperator
@@ -139,6 +140,7 @@ def _get_et_kernel_metadata_from_ops_yaml(ops_yaml_path: str) -> Dict[str, List[
                 ops.append(("aten::" if "::" not in e.get("op") else "") + e.get("op"))
             else:
                 func_entries.append(e)
+        strip_et_fields(es)
         parsed_yaml = parse_native_yaml_struct(
             func_entries, set(), None, path=ops_yaml_path, skip_native_fns_gen=True
         )

kernels/optimized/optimized.yaml

@@ -3,33 +3,41 @@
 # This yaml file contains operators that have optimized kernels available.
 
 - op: _log_softmax.out
-  dispatch:
-    CPU: torch::executor::opt_log_softmax_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_log_softmax_out
 
 - op: bmm.out
-  dispatch:
-    CPU: torch::executor::opt_bmm_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_bmm_out
 
 - op: exp.out
-  dispatch:
-    CPU: torch::executor::opt_exp_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_exp_out
 
 - op: gelu.out
-  dispatch:
-    CPU: torch::executor::opt_gelu_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_gelu_out
 
 - op: le.Scalar_out
-  dispatch:
-    CPU: torch::executor::opt_le_scalar_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_le_scalar_out
 
 - op: le.Tensor_out
-  dispatch:
-    CPU: torch::executor::opt_le_tensor_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_le_tensor_out
 
 - op: native_layer_norm.out
-  dispatch:
-    CPU: torch::executor::opt_native_layer_norm_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_native_layer_norm_out
 
 - op: neg.out
-  dispatch:
-    CPU: torch::executor::opt_neg_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_neg_out

kernels/portable/custom_ops.yaml

@@ -16,8 +16,9 @@
 # TODO(T126667800) Remove dummy_param once custom namespaces are supported in the
 # portable op library.
 - func: allclose.out(Tensor self, Tensor other, float rtol=1e-05, float atol=1e-08, bool equal_nan=False, bool dummy_param=False, *, Tensor(a!) out) -> Tensor(a!)
-  dispatch:
-    CPU: torch::executor::allclose_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::allclose_out
 
 # The argument dummy_param is used solely to disambiguate this op from the native
 # allclose(). Otherwise, code calling this op is identical to the native op:
@@ -26,9 +27,11 @@
 # TODO(T126667800) Remove dummy_param once custom namespaces are supported in the
 # portable op library.
 - func: allclose.Tensor(Tensor self, Tensor other, float rtol=1e-05, float atol=1e-08, bool equal_nan=False, bool dummy_param=False) -> Tensor
-  dispatch:
-    CPU: torch::executor::allclose_tensor
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::allclose_tensor
 
 - func: linear.scratch_example(Tensor input, Tensor weight, Tensor? bias=None, *, Tensor(a!) out, Tensor(b!) _scratch_tensor) -> Tensor(a!)
-  dispatch:
-    CPU: torch::executor::linear_scratch_example
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::linear_scratch_example