rewrite generate_offload_arch script in Python

The change is required to make the script run on Windows. There is
no bash shell on Windows. The alternatives (like MSYS2, Cygwin, etc.)
pollute the build environment with GNU tools and libraries, which
are picked up automatically, making native Windows build impossible.

Author: apwojcik

llvm/lib/OffloadArch/CMakeLists.txt

@@ -1,8 +1,10 @@
+find_package(Python3 REQUIRED COMPONENTS Interpreter)
+
 add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/generated_offload_arch.h
-   COMMAND ${CMAKE_CURRENT_SOURCE_DIR}/make_generated_offload_arch_h.sh ${CMAKE_CURRENT_SOURCE_DIR}
+   COMMAND ${Python3_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/make_generated_offload_arch_h.py ${CMAKE_CURRENT_SOURCE_DIR}
    DEPENDS
-     ${CMAKE_CURRENT_SOURCE_DIR}/make_generated_offload_arch_h.sh
+     ${CMAKE_CURRENT_SOURCE_DIR}/make_generated_offload_arch_h.py
      ${CMAKE_CURRENT_SOURCE_DIR}/amdgpu/pciid2codename.txt
      ${CMAKE_CURRENT_SOURCE_DIR}/amdgpu/codename2offloadarch.txt
      ${CMAKE_CURRENT_SOURCE_DIR}/nvidia/pciid2codename.txt