Install missing VSX headers (POWER) (pytorch#85547)

E.g. `test_cpp_extensions_aot_ninja` fails as it includes `vec.h` which requires the vec/vsx/* headers and `sleef.h`. The latter is also required for AVX512 builds on non MSVC compilers.

Pull Request resolved: pytorch#85547
Approved by: https://github.com/kit1980

aten/src/ATen/CMakeLists.txt

@@ -56,7 +56,7 @@ if(NOT BUILD_CAFFE2 AND NOT BUILD_LITE_INTERPRETER)
   EXCLUDE(ATen_CORE_TEST_SRCS "${ATen_CORE_TEST_SRCS}" ${ATen_CORE_EXCLUDED_TEST_SRCS})
 endif()
 
-file(GLOB base_h "*.h" "detail/*.h" "cpu/*.h" "cpu/vec/vec512/*.h" "cpu/vec/vec256/*.h" "cpu/vec/*.h" "quantized/*.h" "functorch/*.h")
+file(GLOB base_h "*.h" "detail/*.h" "cpu/*.h" "cpu/vec/vec512/*.h" "cpu/vec/vec256/*.h" "cpu/vec/vec256/vsx/*.h" "cpu/vec/*.h" "quantized/*.h" "functorch/*.h")
 file(GLOB base_cpp "*.cpp" "detail/*.cpp" "cpu/*.cpp" "functorch/*.cpp")
 file(GLOB cuda_h "cuda/*.h" "cuda/detail/*.h" "cuda/*.cuh" "cuda/detail/*.cuh")
 file(GLOB cuda_cpp "cuda/*.cpp" "cuda/detail/*.cpp")

setup.py

@@ -1040,6 +1040,7 @@ def main():
         'include/ATen/*.h',
         'include/ATen/cpu/*.h',
         'include/ATen/cpu/vec/vec256/*.h',
+        'include/ATen/cpu/vec/vec256/vsx/*.h',
         'include/ATen/cpu/vec/vec512/*.h',
         'include/ATen/cpu/vec/*.h',
         'include/ATen/core/*.h',
@@ -1148,6 +1149,7 @@ def main():
         'include/THH/*.cuh',
         'include/THH/*.h*',
         'include/THH/generic/*.h',
+        'include/sleef.h',
         "_inductor/codegen/*.h",
         "_inductor/codegen/*.j2",
         'share/cmake/ATen/*.cmake',