Refactor to new raglite toolkit architecture

[mmprotest]

Summary

• migrate the project to a src/ layout with a new raglite package that exposes config, ingestion, embedding, hybrid search, vector backends, and a FastAPI server
• add demo corpus, scripts, CLI commands, documentation, and packaging metadata for the local-first SQLite RAG workflow
• replace the test suite with chunking, embedding, vector backend, and e2e smoke coverage plus CI, pre-commit, and contributor docs

Testing

• pytest -q

---

https://chatgpt.com/codex/tasks/task_e_68e030d1f03483319e488ed4ba9ff016