Fix: Better SBERT error handling + suppress invalid hand-off warnings

Fixed the Pydantic Model Errors for the Token Budget

Fixed Agent Reasoning and Semanttic Layer Integration issues

feat: Add 4 major system improvements - semantic layer, error recovery, token budget, parallel execution

Fix Phase 1 errors: schema extraction and message type handling

Refactor: Move workflow context out of LLM prompts into structured state

feat: Initial commit - Data Science Agent with React frontend and FastAPI backend