Two Scores per Degree

A plant's height over a season rises, levels off, then drops as it wilts. You have some measurements split into training points and held-back test points.

Write scores_by_degree(x_train, y_train, x_test, y_test, degrees).

For each degree in degrees, in order:

• Fit make_pipeline(PolynomialFeatures(degree), LinearRegression()) on the training points, with x as the only feature.
• Score it on the training points, then on the test points (score gives R²).

Return a list with one [degree, train_score, test_score] per degree. Round both scores to 3 decimal places.