CMake: Include instead of copying cpu kernel files (pytorch#67656)

Summary:
Pull Request resolved: pytorch#67656

Currently, each cpu kernel file is copied into the build folder 3 times to give them different compilation flags. This changes it to instead generate 3 files that `#include` the original file. The biggest difference is that updating a copied file requires `cmake` to re-run, whereas include dependencies are natively handled by `ninja`.

A side benefit is that included files show up directly in the build dependency graph, whereas `cmake` file copies don't.

Test Plan: Imported from OSS

Reviewed By: dagitses

Differential Revision: D32566108

Pulled By: malfet

fbshipit-source-id: ae75368fede37e7ca03be6ade3d4e4a63479440d

cmake/Codegen.cmake

@@ -67,7 +67,7 @@ if(INTERN_BUILD_ATEN_OPS)
     set_source_files_properties(${CMAKE_CURRENT_LIST_DIR}/../aten/src/ATen/MapAllocator.cpp PROPERTIES COMPILE_FLAGS "-fno-openmp")
   endif()
 
-  file(GLOB cpu_kernel_cpp_in "${CMAKE_CURRENT_LIST_DIR}/../aten/src/ATen/native/cpu/*.cpp" "${CMAKE_CURRENT_LIST_DIR}/../aten/src/ATen/native/quantized/cpu/kernels/*.cpp")
+  file(GLOB cpu_kernel_cpp_in "${PROJECT_SOURCE_DIR}/aten/src/ATen/native/cpu/*.cpp" "${PROJECT_SOURCE_DIR}/aten/src/ATen/native/quantized/cpu/kernels/*.cpp")
 
   list(APPEND CPU_CAPABILITY_NAMES "DEFAULT")
   list(APPEND CPU_CAPABILITY_FLAGS "${OPT_FLAG}")
@@ -129,10 +129,10 @@ if(INTERN_BUILD_ATEN_OPS)
   # See NOTE [ Linking AVX and non-AVX files ]
   foreach(i RANGE ${NUM_CPU_CAPABILITY_NAMES})
     foreach(IMPL ${cpu_kernel_cpp_in})
-      string(REPLACE "${CMAKE_CURRENT_LIST_DIR}/../aten/src/ATen/" "" NAME ${IMPL})
+      file(RELATIVE_PATH NAME "${PROJECT_SOURCE_DIR}/aten/src/ATen/" "${IMPL}")
       list(GET CPU_CAPABILITY_NAMES ${i} CPU_CAPABILITY)
       set(NEW_IMPL ${CMAKE_BINARY_DIR}/aten/src/ATen/${NAME}.${CPU_CAPABILITY}.cpp)
-      configure_file(${IMPL} ${NEW_IMPL} COPYONLY)
+      configure_file("${PROJECT_SOURCE_DIR}/cmake/IncludeSource.cpp.in" ${NEW_IMPL})
       set(cpu_kernel_cpp ${NEW_IMPL} ${cpu_kernel_cpp}) # Create list of copies
       list(GET CPU_CAPABILITY_FLAGS ${i} FLAGS)
       if(MSVC)

cmake/IncludeSource.cpp.in

@@ -0,0 +1 @@
+#include "@IMPL@"