Feature/stem ai integration

[halinhtvn3a]

This pull request introduces improvements to the context gathering and ingestion flow for classroom data, adds a new test endpoint for triggering progress events, and enhances logging and debugging for RAG (Retrieval Augmented Generation) pipelines. It also updates protocol buffer definitions to support richer student progress summaries. The most significant changes are grouped below:

RAG Ingestion & Context Gathering Improvements:

• Added an IngestionService provider and dependency injection into relevant services and event handlers, enabling debounced, event-driven ingestion of classroom data for RAG pipelines (app/api/http/dependencies.py). [1] [2] [3]
• Enhanced the context gathering logic to prioritize ingestion documents, combine results from general and ingestion-focused retrieval, and provide detailed logging and fallbacks when ingestion data is missing (app/core/context/gather.py).
• Improved logging in the context builder to provide more transparency at each stage of the context construction pipeline (app/core/context/builder.py).

API & Testing Enhancements:

• Added a new /test/trigger-progress-event endpoint and request model to allow manual triggering of classroom student progress events, supporting both direct and RabbitMQ-based event handling for testing the ingestion and snapshot update flow (app/api/http/routers/teacher.py). [1] [2]

Protocol Buffer Updates:

• Introduced the GrpcStudentProgressSummary message and added a repeated field for student progress summaries to the classroom snapshot response, allowing richer progress reporting (classrooms.proto). [1] [2]

Agent Logging Cleanup:

• Removed verbose logging of full prompts in the plan-solve agent to reduce log noise and potential information leakage (app/core/agent/plan_solve_agent.py). [1] [2] [3]