Invoke Agent wrapper span populates tool-call turns and gen_ai.tool.definitions

[rads-1996]

Fixes #172

[Copilot]

Pull request overview

This PR addresses issue #172 by improving LangChain invoke_agent span aggregation so wrapper spans can carry ordered ReAct-style message history and tool definitions.

Changes:

• Adds LangChain utilities for tool-call response extraction, role normalization, tuple shorthand handling, and agent-history building.
• Updates agent span aggregation to build cumulative input history from LLM/tool child runs and defer final assistant output until finalization.
• Updates and expands LangChain tests for spec-compliant message extraction and aggregation behavior.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
src/microsoft/opentelemetry/_genai/_langchain/_utils.py	Adds message/tool-response helpers and improves role/message extraction.
src/microsoft/opentelemetry/_genai/_langchain/_tracer.py	Aggregates invoke-agent input history, final output messages, and tool definitions.
tests/langchain/test_utils.py	Updates session assertions and adds structured message extraction tests.
tests/langchain/test_tracer.py	Updates tool-output behavior tests and adds aggregation/tool-definition tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.