Feature/knowledge agent

[amrit110]

This pull request introduces a new, well-structured agent evaluation and knowledge-grounded QA agent framework, with a focus on modularity, configuration, and rich display utilities. It adds core components for knowledge-grounded QA using Google ADK and Google Search, provides rich output formatting for Jupyter notebooks, and centralizes configuration management. Additionally, it cleans up placeholder/example code and updates development environment and pre-commit configurations.

Agent Evaluation and Knowledge-Grounded QA Framework

• Added a new aieng.agent_evals.knowledge_agent package implementing a knowledge-grounded QA agent using Google ADK and Google Search, including agent, configuration, evaluation, grounding tool, and session management modules. [1] [2] [3]
• Introduced GroundedResponse and GroundingChunk models for structured agent outputs, and a utility to format responses with inline citations.
• Centralized configuration using Pydantic settings, supporting environment variables and .env file loading.

Rich Display Utilities

• Added aieng.agent_evals.display module with reusable functions for displaying agent responses, evaluation results, metrics, and messages using the rich library, with special support for Jupyter notebooks.
• Exposed display utilities in the aieng.agent_evals package for easy import and use.

Cleanup and Environment Improvements

• Removed placeholder/example implementation modules and their references for a cleaner codebase. [1] [2]
• Updated .env.example to reflect new agent configuration variables, including Gemini/OpenAI API settings and Langfuse tracing options.
• Improved pre-commit configuration: changed the Ruff hook to ruff-check and expanded ignored error codes for nbqa-ruff. [1] [2]