feat: add comprehensive integration tests for transport providers

[Pratham-Mishra04]

Add Bifrost Integration Tests for Transport Implementations

This PR adds a comprehensive end-to-end test suite for validating AI integrations through the Bifrost proxy. The test framework provides uniform testing across multiple AI providers including OpenAI, Anthropic, Google, LiteLLM, LangChain, LangGraph, and Mistral.

Key Features

• Centralized configuration system with YAML-based settings and environment variable support
• Integration-specific test implementations for OpenAI, Anthropic, Google, and LiteLLM
• 11 standardized test categories covering chat, tool calling, image processing, and multimodal workflows
• Flexible test execution with command-line options for selective testing
• Robust error handling with graceful fallbacks and detailed reporting
• Support for parallel test execution and verbose output modes

Implementation Details

• Added .gitignore entries for Python virtual environments and cache files
• Created a Makefile with convenient commands for running tests and managing the environment
• Implemented test runners for both sequential and parallel execution
• Added comprehensive documentation in README.md with examples and troubleshooting guides
• Developed integration-specific client implementations with proper error handling
• Created shared test utilities for consistent assertions and data handling

The test suite is designed to be extensible, allowing easy addition of new integrations while maintaining consistent test coverage across all providers.