Add comprehensive GitHub Copilot instructions for RAP repository

[Copilot]

This PR adds a comprehensive .github/copilot-instructions.md file that provides GitHub Copilot coding agents with detailed guidance on working with the RAP (Risk-Aware Prediction) repository. The instructions cover all aspects of development workflow from initial setup to validation.

Key Features

Installation and Environment Setup

• Step-by-step virtual environment creation and dependency installation
• Network connectivity issue warnings with fallback strategies
• Specific timeout recommendations (30+ min for install, 90+ min for training)
• Documentation of common installation failures in restricted environments

Validated Commands
Every command in the instructions has been tested and verified:

• Package structure validation: PYTHONPATH=/home/runner/work/RAP/RAP python3 -c "import risk_biased"
• Configuration loading tests: from risk_biased.config import learning_config
• Syntax checking: python3 -m py_compile risk_biased/models/*.py
• Code formatting with yapf

Comprehensive Project Structure

• Complete directory listings with actual file sizes and permissions
• Configuration file purposes and requirements (especially paths.py setup)
• Training scenarios: didactic simulation vs Waymo Open Motion Dataset
• Dependency overview with exact version requirements

User Validation Scenarios
Three complete validation workflows:

1. Basic Setup Validation: Package import, config files, directory structure
2. Training Pipeline: Virtual environment setup, installation, quick training test
3. Code Quality: Syntax checking, formatting, import validation after changes

Environment Limitations Documentation

• Externally-managed Python environments and virtual environment requirements
• CUDA dependencies and GPU memory considerations
• Two-phase training complexity and potential GPU OOM issues
• WandB integration requirements and checkpoint management

The instructions start with a firm directive to "Always reference these instructions first and fallback to search or bash commands only when you encounter unexpected information" as requested, ensuring Copilot agents prioritize the validated guidance over exploration.

The file also includes realistic timing expectations for all operations, with explicit "NEVER CANCEL" warnings for long-running builds and training processes, helping prevent premature termination of critical operations.

Fixes #4.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.