Description Set up Vitest, Testing Library, and jsdom. Fetch one real MET response, trim it to a few timeseries entries, and commit it as a fixture. Mock metClient's fetchCompact (not raw fetch) in any test that needs data. Cover a snapshot test, a props test, a state test, an interaction test, and a query-integration test. Rationale Rubric requires Vitest, a snapshot test, and tests of props/state/user interaction, with mocking so tests don't hit the real network. Using a real (trimmed) API response rather than a hand-written fixture is what actually surfaces which fields MET sometimes omits (e.g. wind gust), instead of us guessing.