Vectorize RandomTranslation

[james77777778]

What does this PR do?

Related to #291

Benchmark results:

	Mode	n=100	n=200	n=500	n=1000
OldRandomTranslation	Eager	0.77s	1.69s	7.84s	30.13s
OldRandomTranslation	Graph	0.33s	1.19s	7.41s	29.92s
RandomTranslation	Eager	0.0072s	0.0069s	0.0087s	0.0119s
RandomTranslation	Graph	0.0014s	0.0019s	0.0042s	0.0077s

OldRandomTranslation is too slow that I need to set num_images = [100, 200, 500, 1000] for benchmarks

Both implementations fail to run with XLA due to lacking ImageProjectiveTransformV3 support.
tensorflow/tensorflow#55194

Similar to #1435, I add a test to check random translation is applied for each image.

    def test_random_translation_on_batched_images_independently(self):
        image = tf.random.uniform(shape=(100, 100, 3))
        input_images = tf.stack([image, image], axis=0)

        layer = preprocessing.RandomTranslation(
            height_factor=0.5, width_factor=0.5
        )

        results = layer(input_images)
        self.assertNotAllClose(results[0], results[1])

Also, I found that RandomTranslation is missing in keras_cv/layers/preprocessing/ragged_image_test.py and keras_cv/layers/preprocessing/with_labels_test.py. I added it to both tests and they passed on my local env.

    (
        "RandomTranslation",
        layers.RandomTranslation,
        {"height_factor": 0.5, "width_factor": 0.5},
    ),

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue? Please add a link
  to it if that's the case.
• Did you write any new necessary tests?
• If this adds a new model, can you run a few training steps on TPU in Colab to ensure that no XLA incompatible OP are used?

Who can review?

@LukeWood @ianstenbit

[LukeWood]

Wow - great PR. I'm pretty surprised by just how large the performance delta here is between the vectorized/unvectorized version is in this case (the others seem to be much smaller). Great work!

[ianstenbit]

Awesome result -- thanks for the PR!

[ianstenbit]

/gcbrun