[Workaround] Failed to load shared library (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64'))

This issue comes up on some Mac M series environments where cmake builds the llama.cpp shared library for the wrong architecture. The solution is to reinstall llama-cpp-python with:

```
CMAKE_ARGS="-DCMAKE_OSX_ARCHITECTURES=arm64" pip install --upgrade --verbose --force-reinstall --no-cache-dir llama-cpp-python
```

If anyone has a good way to automatically detect this in the CMakeLists.txt please let me know.