feat: sequence check vllm

[kjnasdfkjnsadfkjnds]

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS ABOVE HAVE BEEN CONSIDERED.

Purpose

Add run_seed field to chat completion responses to enable sequence validation in decentralized inference
systems. This implements the foundation for Stage 1 sequence check validation as described in the inference
validation proposal.

When a user provides a seed parameter, vLLM now:

1. Generates a deterministic run_seed by hashing user_seed + inference_id
2. Uses the first 16 hex characters of run_seed to derive the actual sampling seed
3. Returns run_seed in the logprobs field of the response

This allows validators to verify that generated tokens were sampled from the claimed model's probability
distribution.

Test Plan

# Run validation utils tests
pytest tests/entrypoints/openai/test_validation_utils.py -v

# Test with actual inference
curl -X POST http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "facebook/opt-125m",
    "messages": [{"role": "user", "content": "Count to 5"}],
    "seed": 42,
    "logprobs": true,
    "top_logprobs": 5,
    "max_tokens": 20,
    "chat_template": "{{messages[0].content}}"
  }'

Test Result

All tests pass:

tests/entrypoints/openai/test_validation_utils.py::test_generate_run_seed PASSED
tests/entrypoints/openai/test_validation_utils.py::test_generate_run_seed_deterministic PASSED
tests/entrypoints/openai/test_validation_utils.py::test_generate_run_seed_different_inference_ids PASSED
tests/entrypoints/openai/test_validation_utils.py::test_generate_run_seed_none PASSED
tests/entrypoints/openai/test_validation_utils.py::test_compute_derived_seed PASSED
tests/entrypoints/openai/test_validation_utils.py::test_compute_derived_seed_deterministic PASSED
tests/entrypoints/openai/test_validation_utils.py::test_compute_derived_seed_none PASSED

Response now includes run_seed:

{
  "choices": [{
    "logprobs": {
      "content": [...],
      "run_seed": "4ef75adbd0ec3fba28ade0cf79ba5d4afea742e476567517d348836bb1876c9b"
    }
  }]
}

(Optional) Documentation Update

None required, this is an additive change that adds an optional field to existing responses

BEFORE SUBMITTING, PLEASE READ https://docs.vllm.ai/en/latest/contributing (anything written below this line will be removed by GitHub Actions)