Update on "Reuse GELU implementation from PyTorch core"

kernels/optimized doesn't need to support embedded systems, so it can just take a header-only dep on PyTorch.

Note that, because we will pick up Sleef internally and ignore it
externally thanks to ATen vec, this PR gets to enable optimized GELU in OSS.

Testing: CI to make sure this doesn't break mobile build modes; happy to take advice on anything not currently covered that might break.

Differential Revision: [D66335522](https://our.internmc.facebook.com/intern/diff/D66335522/)

[ghstack-poisoned]

Author: swolchok

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-2ea4b56ec872424e486c4fe2d55da061067a2ed3
+f0c50a619c282a01ef1ad33690b7ab5e615d9fef

configurations/CMakeLists.txt

@@ -42,9 +42,8 @@ if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
 
   generate_bindings_for_kernels(
     LIB_NAME "optimized_native_cpu_ops_lib" FUNCTIONS_YAML
-    ${CMAKE_CURRENT_BINARY_DIR}/merged.yaml
+    ${CMAKE_CURRENT_BINARY_DIR}/merged.yaml ADD_EXCEPTION_BOUNDARY
   )
-  target_compile_options(optimized_native_cpu_ops_lib PRIVATE "-fexceptions")
   message("Generated files ${gen_command_sources}")
 
   # optimized_native_cpu_ops_lib: Register optimized op kernels into the runtime
@@ -57,6 +56,7 @@ if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
     DEPS
     executorch
   )
+  target_compile_options(optimized_native_cpu_ops_lib PRIVATE "-fexceptions")
 
   install(TARGETS optimized_native_cpu_ops_lib DESTINATION lib)
 endif()