Updates to ragged tensor guide

* Added section on using RaggedTensors with TF APIs (Keras, tf.Example, Datasets, tf.function, and SavedModel).

* Updated to reflect the fact that any combination of ragged & uniform dimensions is now supported.

* Updated section on RaggedTensor encoding.

* Simplified section on Evaluating ragged tensors, and added information about `RaggedTensor.numpy()`.

* Updated section on Ragged vs Sparse.

* Moved description of `ragged_rank` and `flat_values` to the "RaggedTensor encoding" section.

PiperOrigin-RevId: 311852094

Author: edloper