Using Noir with vLLM

Integrate Noir with vLLM for high-performance local LLM inference.

Serve a model locally with vLLM and point Noir at it.

Setup

Install vLLM: Follow the official installation guide.
Serve a Model:
```
vllm serve google/gemma-4-31B-it
```
This starts an OpenAI-compatible server on port 8000.

Usage

noir scan ./spec/functional_test/fixtures/hahwul \
     --ai-provider=vllm \
     --ai-model=google/gemma-4-31B-it

The vllm preset points at http://localhost:8000/v1. For another host or port, pass the URL instead, for example --ai-provider=http://gpu-box:8000/v1.