Cut the Table in Two

A gym keeps a table about its members. One column is the thing it wants to predict. All the other columns are clues.

Task: write predict_from_table(table, label_column, new_rows).

• table is a dict that maps each column name to a list of values. All lists have the same length. It includes the label column, and the label column is not always the last one.
• label_column is the name of the column to predict.
• new_rows is a dict shaped like table but without the label column. Its feature columns are in the same order as in table.
• Build X from every column except the label, and y from the label column.
• Train a KNeighborsClassifier(n_neighbors=3) on X and y, then predict new_rows.
• Return the predictions as a plain list, one per new row.