add a Dockerfile for nano-vllm

[pacoxu]

This can help users run nano-vllm using docker.

docker build -t nano-vllm:local .

docker run --gpus all --rm -it \
  --shm-size=2g \
  -v ~/huggingface/Qwen3-0.6B:/models/Qwen3-0.6B \
  nano-vllm:local \
  python example.py

Previously, this was mentioned in #56.