feat: added nilai prod configs

[jcabrero]

This PR adds the nilai f910 endpoint configuration for models

[Copilot]

Pull Request Overview

This PR adds production configuration for the nilai endpoint by introducing a new Docker Compose file that defines two GPU-enabled services for running LLM models. The configuration supports both a Llama 3.1-8B model and a GPT-OSS-20B model with specific GPU memory allocation and performance optimizations.

• Adds GPU-enabled service configurations for Llama 3.1-8B and GPT-OSS-20B models
• Configures health checks and service dependencies for proper startup ordering
• Sets up shared volume for Hugging Face model caching

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}