Prevent long context

[jmsevin]

This pull request introduces middleware enhancements to the agent creation process, improves test coverage for middleware integration, and makes several minor code cleanups and import fixes.

Agent Middleware Integration:

• Added SummarizationMiddleware to the agent creation in _create_agent, ensuring that long conversations are summarized automatically when token counts exceed 64,000. [1] [2]

Testing Enhancements:

• Added a new test test_create_agent_adds_summarization_middleware to verify that SummarizationMiddleware is correctly added to the agent's middleware stack.

Code Quality and Maintenance:

• Cleaned up and reordered imports in src/app/bibliography/api/bibliography.py and src/app/bibliography/helpers/helpers.py for consistency and clarity. [1] [2]
• Refactored imports in src/app/tests/services/test_helpers.py to group related functionality and remove redundant code. [1] [2]

Configuration:

• Disabled LangSmith tracing for the test environment by adding LANGSMITH_TRACING_ENABLED=false to pytest.ini.