feat: add AG2 target for evaluating AG2/AutoGen agents

[faridun-ag2]

Summary

• Adds AG2 (formerly AutoGen) as a new target type, enabling evaluation of AG2-based agents through the agent-evaluation framework
• First non-AWS target, demonstrating the framework's extensibility
• AG2 is an open-source multi-agent framework with 500K+ monthly PyPI downloads and 4,300+ GitHub stars

What's included

• src/agenteval/targets/ag2/ — AG2Target implementation extending BaseTarget
• Unit tests with full mocking of AG2 components (6 tests)
• Documentation at docs/targets/ag2.md
• AG2 dependency added to requirements.txt (ag2[openai]>=0.11.4,<1.0)
• CHANGELOG.md updated under Unreleased

Configuration Example

target:
  type: ag2
  model: gpt-4o-mini
  api_type: openai
  system_message: "You are a helpful assistant."
  max_turns: 3

Test Plan

• black formatting passes
• isort passes
• flake8 passes
• All unit tests pass (pytest tests/src/agenteval/targets/ag2/)
• Full test suite passes (pytest tests/ — 80/80)
• AG2Target.invoke() returns TargetResponse correctly
• Agent reset between invocations works
• Documentation follows existing target docs pattern
• CHANGELOG.md updated
• No existing targets/tests modified

[faridun-ag2]

Hi @bobbywlindsey @sharonxiaohanli — would appreciate your review on this PR. It adds AG2 (formerly AutoGen) as the first non-AWS target, demonstrating the framework's extensibility. Happy to address any feedback!