[Misc] Separate hf dataset sampling function from benchmark_serving.py

[Isotr0py]

• To support different VLM datasets, the sampling function is growing large and complex in benchmark_serving.py
• This PR aims to separate and decouple dataset sampling function from benchmark_serving.py to keep benchmark script clean

[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.

🚀

[ywang96]

Sorry I missed this PR.

Overall I think if we're going to create a separate datatset_sample_func.py file, then we should consolidate all dataset sampling functions in there (ShareGPT, Sonnet, Random, etc).

Absolutely no rush if you have other priorities to focus on.

[Isotr0py]

@ywang96 I have migrated all datasets from benchmark_serving.py to datatset_sample_func.py, PTAL!

[ywang96]

:O Sorry I missed this PR!

@Shaoting-Feng Since you're also interested in working on this, maybe you can take a first pass (and try to see how to integrate this with benchmark_throughput.py)!