[Hardware][Metal] Apple Metal support

[skyzh]

fix #2081

This patch makes some parts of vllm run with Apple Metal by using the PyTorch MPS fallback mode (see build_and_run.sh), which ensures that PyTorch operators can run natively on MPS while other operations run on CPU. Though the framework runs end-to-end and produces texts, it's not producing a sensible result:

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "Qwen/Qwen2.5-0.5B-Instruct",
        "prompt": "San Francisco is a",
        "max_tokens": 7,
        "temperature": 0
    }'
{"id":"cmpl-aba9771f627e400fab7be25d8b310fd5","object":"text_completion","created":1738386309,"model":"Qwen/Qwen2.5-0.5B-Instruct","choices":[{"index":0,"text":"!!!!!!!","logprobs":null,"finish_reason":"length","stop_reason":null,"prompt_logprobs":null}],"usage":{"prompt_tokens":4,"total_tokens":11,"completion_tokens":7,"prompt_tokens_details":null}}%

...which needs further debugging, and comments welcomed -- I have no idea what's going on there.

For a full Metal support, we would have to implement all current CUDA kernels with Metal, which will be a lot of work. So this patch is the very first step before we have full Metal support.

In general, the patch assumes the Metal platform is a CPU-based platform (i.e., using CPU workers) with PyTorch MPS backend. This can also be improved in the future, for example, using the GPU scheduler for Metal.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀