Schema Validation + Smoke Tests

# Schema Validation + Smoke Tests Write 5 smoke tests including bad-input cases for the FastAPI serving lesson. ## 1. Contract first `/health` returns 200 + version. `/predict` validates input via pydantic schemas (types, ranges, required fields) and returns predictions + model version. Unknown fields are rejected or explicitly ignored — pick one and log it. ## 2. The 5 smoke tests 1. `/health` 200 with version string. 2. Valid `/predict` returns 200 with the right schema. 3. Missing required field returns 422. 4. Out-of-range value returns 422. 5. Wrong type (string for float) returns 422. Each test asserts status + a key field, never the full body. ## 3. Extend the backend pattern Mirror `backend/app/main.py` `/health` shape so the serving code reads like the codebase it deploys into. Smoke file lives at `tests/test_smoke.py` and runs inside the Docker smoke command from m0.