[Docs][ROCm] Detailed instructions to build from source

[WoosukKwon]

For some environments like Runpod, users can't use docker and have to build vLLM from source. This PR adds the detailed instructions for it, which are extracted from Dockerfile.rocm.

cc @hongxiayang @mawong-amd

[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 consists a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of default ones by unblocking the steps in your fast-check build on Buildkite UI.

Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge).

To run full CI, you can do one of these:

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

🚀

[simon-mo]

@pandyamarut ^

[WoosukKwon]

@pandyamarut I wrote this script to test llama models on a 8xMI300x instance at Runpod. It worked with the ROCm 6.1 PyTorch image at Runpod.

[pandyamarut]

@simon-mo @WoosukKwon This looks good.

[hongxiayang]

Thank you @WoosukKwon ! Appreciate it.