Added a vllm quickstart playbook and scripts for curl and gradio web interactive client

[hongxiayang]

In this playbook, you will learn how to:

• Set up and run vLLM with ROCm support using Docker for high-performance LLM inference on AMD GPUs
• Download and configure language models from Hugging Face for use with vLLM
• Start and configure a vLLM server with OpenAI-compatible API endpoints on port 8000
• Test the server using curl commands and API requests
• Launch and use the Gradio web interface (port 7860) for interactive chat with real-time streaming responses
• Configure server parameters like GPU memory utilization, model length limits, and multi-GPU support
• Make API calls to the vLLM server using both streaming and non-streaming requests
• Troubleshoot common issues with server startup, memory, and client connections

[danielholanda]

@hongxiayang Please fix the CI failure by moving the additional scripts you added to an assets folder inside the vllm-inference repo.

[danielholanda]

Next steps:

• Daniel to investigate alternatives on preinstalling vLLM or facilitating installs without Docker.

[hongxiayang]

@hongxiayang Please fix the CI failure by moving the additional scripts you added to an assets folder inside the vllm-inference repo.

@danielholanda Done.

[danielholanda]

Thanks for your contribution @hongxiayang . Eddie will now take it over and adapt it to work with vllm whls instead of docker.