Add NVIDIA TensorRT-LLM optimization guide for GPT-OSS models

[jayrodge]

Summary

Adds a comprehensive guide for optimizing OpenAI GPT-OSS models using NVIDIA TensorRT-LLM.

Changes

• Add detailed guide for optimizing gpt-oss-20b and gpt-oss-120b models
• Include hardware prerequisites (16GB+ VRAM, recommended GPUs)
• Provide installation instructions for TensorRT-LLM via NGC and Docker
• Add Python API examples for model loading and inference
• Include performance optimization tips and next steps

Benefits

• Helps users optimize GPT-OSS models for high-performance inference
• Provides clear hardware requirements and setup instructions
• Includes practical code examples for immediate use