Fix lint warning and import error in data_types_and_io tf example

[eapolinario]

Unfortunately the example added in #1739 cannot make use of the is_container ImageSpec functionality because some of the imported types are part of the task definition, e.g.:

@task
def train_model() -> tf.keras.Model:
    model = tf.keras.Sequential(
        [tf.keras.layers.Dense(128, activation="relu"), tf.keras.layers.Dense(10, activation="softmax")]
    )
    model.compile(optimizer="adam", loss="sparse_categorical_crossentropy", metrics=["accuracy"])
    return model

Notice the use of tf.keras.Model as the task output.

This PR removes the is_container check and fixes the lint warning.