Add ray[default] to wget to run distributed inference out of box

[Jeffwan]

part of #11137

Why do we need such PR?

1. User wants to leverage vllm default image to manage RayCluster version instead of managing separate ray images and vLLM distributions which frequently surface version compatibility problems
2. ray[default] package provides more capabilities like submission api and dashboard which is essential to run vLLM distributed inference in cluster environment
3. wget is used in kubeRay prob.https://github.com/ray-project/kuberay/blob/e595ee4c6297fb6b385421f7ca34fbd7c1c0b49f/ray-operator/controllers/ray/common/pod.go#L253 and https://github.com/ray-project/kuberay/blob/e595ee4c6297fb6b385421f7ca34fbd7c1c0b49f/ray-operator/controllers/ray/utils/constant.go#L178. This can be definitely changed to curl but I feel adding a new package won't take that many room here.
4. Adding wget and ray[default] won't be risky. I notice some other files already include this change.

• vllm/requirements-tpu.txt

  Line 11 in f9ecbb1

  ray[default]

• vllm/Dockerfile.arm

  Line 11 in f9ecbb1

  && apt-get install -y curl ccache git wget vim numactl gcc-12 g++-12 python3 python3-pip libtcmalloc-minimal4 libnuma-dev \

FIX

[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.

🚀

[Jeffwan]

this change is not directly related to above issue. I did verify the testing on my end and it's working fine.

vllm/tests/entrypoints/openai/test_chat.py

Lines 469 to 503 in 2d1b9ba

	async def test_guided_choice_chat(client: openai.AsyncOpenAI,
	guided_decoding_backend: str,
	sample_guided_choice):
	messages = [{
	"role": "system",
	"content": "you are a helpful assistant"
	}, {
	"role":
	"user",
	"content":
	"The best language for type-safe systems programming is "
	}]
	chat_completion = await client.chat.completions.create(
	model=MODEL_NAME,
	messages=messages,
	max_completion_tokens=10,
	extra_body=dict(guided_choice=sample_guided_choice,
	guided_decoding_backend=guided_decoding_backend))
	choice1 = chat_completion.choices[0].message.content
	assert choice1 in sample_guided_choice
	messages.append({"role": "assistant", "content": choice1})
	messages.append({
	"role": "user",
	"content": "I disagree, pick another one"
	})
	chat_completion = await client.chat.completions.create(
	model=MODEL_NAME,
	messages=messages,
	max_completion_tokens=10,
	extra_body=dict(guided_choice=sample_guided_choice,
	guided_decoding_backend=guided_decoding_backend))
	choice2 = chat_completion.choices[0].message.content
	assert choice2 in sample_guided_choice
	assert choice1 != choice2