Update and consolidate Keras and TensorFlow taskrunner openfl-workspaces under a single TensorFlowTaskRunner class

[kta-intel]

The current TensorFlow TaskRunner is outdated and currently only supports 1.x. The current Keras TaskRunner is also outdated and only supports TF implementation.

Since stock Keras has support for multiple frameworks (i.e. torch, tf, and JAX), and the recommended practice by the TF team is to use Keras by default with TF, the current keras and tensorflow taskrunner openfl-workspaces should be updated and consolidated under a single tensorflow taskrunner class

This will require:

• Updating necessary openfl.federated.task files (i.e. updating KerasTaskRunner to TensorFlowTaskrunner and removing/deprecating the original TensorFlowTaskrunner)
• Updating workspaces (tf_* and keras_* workspaces should run on the new taskrunner, keras_* workspaces should update naming convention to tf)
• Fixing how optimizer weights are being updated because legacy optimizer is no longer be supported in the latest keras