[CI/Build] simplify Dockerfile build for ARM64 / GH200

[cennn]

From PR: 10499
Fix Issue: 2021

This contribution focuses on simplifying the Dockerfile build process for ARM64 systems. Unnecessary build from source has been removed and requirements handling has been optimized to ensure the correct installation of torch and bitsandbytes for ARM64+CUDA compatibility. The changes have been tested on the Nvidia GH200 platform with models meta-llama/Llama-3.1-8B and Qwen/Qwen2.5-0.5B-Instruct

The following command was used to build and confirmed working on Nvidia GH200:
docker build . --target vllm-openai --platform "linux/arm64" -t cenncenn/vllm-gh200-openai:v0.6.4.post1 --build-arg max_jobs=66 --build-arg nvcc_threads=2 --build-arg torch_cuda_arch_list="9.0+PTX" --build-arg vllm_fa_cmake_gpu_arches="90-real" --build-arg RUN_WHEEL_CHECK='false'

[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.

🚀

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @cennn.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork