[Serve.llm] feat: add missing repetition_penalty vLLM sampling param

[ArthurBook]

@kouroshHakha

Why are these changes needed?

repetition_penality is a valid vLLM SamplingParam but is missing from the ray.serve counterpart. This causes it to be ignored when passed!

Related issue number

Closes #53221

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
  • I've added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[lk-chen]

Thanks @ArthurBook for the contribution. I think you need to update some tests e.g.

ray/python/ray/llm/tests/serve/cpu/deployments/llm/multiplex/test_multiplex_deployment.py

Lines 78 to 80 in efd333a

	"response_format": None,
	"top_k": None,
	"seed": None,

[ArthurBook]

Thanks! Pushed. This made me realize that the same thing may be happening to other params listed under extra SamplingParams? I don't see e.g. use_beam_search being passed to vLLM anywhere? LMK and I'll add those too. EDIT: Looks like use_beam_search has been dropped all together from SamplingParams.

[kouroshHakha]

Thanks a ton @ArthurBook . LGTM