[reland][Inductor] Rename cpp_wrapper_cuda.py as cpp_wrapper_gpu.py (#136046)

Summary: Reland https://github.com/pytorch/pytorch/pull/135313 after fixing internal build issues

Test Plan: CI

Differential Revision: D62658837

Pull Request resolved: https://github.com/pytorch/pytorch/pull/136046
Approved by: https://github.com/chenyang78, https://github.com/etaf, https://github.com/jansel

tools/amd_build/build_amd.py

@@ -207,7 +207,7 @@ hipify_python.hipify(
     ignores=ignores,
     extra_files=[
         "torch/_inductor/codegen/cpp_wrapper_cpu.py",
-        "torch/_inductor/codegen/cpp_wrapper_cuda.py",
+        "torch/_inductor/codegen/cpp_wrapper_gpu.py",
         "torch/_inductor/codegen/wrapper.py",
     ],
     out_of_place_only=args.out_of_place_only,

torch/_inductor/codegen/common.py

@@ -232,7 +232,7 @@ def get_wrapper_codegen_for_device(device: str, cpp_wrapper: bool = False):
 def init_backend_registration():
     from .cpp import CppScheduling
     from .cpp_wrapper_cpu import CppWrapperCpu
-    from .cpp_wrapper_cuda import CppWrapperGpu
+    from .cpp_wrapper_gpu import CppWrapperGpu
     from .cuda_combined_scheduling import CUDACombinedScheduling
     from .halide import HalideScheduling
     from .triton import TritonScheduling