Apply @tf.function to the __call__ method of RandAugment

[LukeWood]

No description provided.

[bhack]

Probably it is better to have some interface to control @tf.function in the base class: https://github.com/keras-team/keras-cv/blob/master/keras_cv/layers/preprocessing/random_augmentation_pipeline.py

So that we could conditionally control also the pipeline compilation similar to keras-team/keras-io#1541

[LukeWood]

Removed from v0.1 because MOST users will use this in a tf.data pipeline or keras model that automatically does this.

[LukeWood]

I do agree a higher level abstraction would be nice though @bhack

[LukeWood]

Closing as tf.data performs this, or the model itself.