feat: install Run:ai model streamer for vllm (#4848)

Signed-off-by: Guangtong Bai <guangtong.bai@gmail.com>

container/deps/vllm/install_vllm.sh

@@ -126,7 +126,7 @@ git checkout $VLLM_REF
 echo "\n=== Installing vLLM & FlashInfer ==="
 echo "Installing vLLM $VLLM_REF from PyPI..."
 
-uv pip install vllm[flashinfer]==$VLLM_REF --torch-backend=${TORCH_BACKEND}
+uv pip install vllm[flashinfer,runai]==$VLLM_REF --torch-backend=${TORCH_BACKEND}
 uv pip install flashinfer-cubin==$FLASHINF_REF
 uv pip install flashinfer-jit-cache==$FLASHINF_REF --extra-index-url https://flashinfer.ai/whl/${TORCH_BACKEND}
 

pyproject.toml

@@ -56,7 +56,7 @@ trtllm =[
 vllm = [
     "uvloop",
     "nixl[cu12]<=0.7.1",
-    "vllm[flashinfer]==0.12.0",
+    "vllm[flashinfer,runai]==0.12.0",
 ]
 
 sglang = [