[Task]: Support Keras 3 in TFModelHandlers

[riteshghorse]

What needs to happen?

Right now, if the models are not saved in .keras format, keras 3 will throw an error as observed in #30613. Allow loading both kind of models (models saved in SavedModel format and .keras) in TFModelHandlers.

Remove the keras<3.0.0 dependency from here once the support is added.

Issue Priority

Priority: 2 (default / most normal work should be filed as P2)

Issue Components

• [X] Component: Python SDK
• [ ] Component: Java SDK
• [ ] Component: Go SDK
• [ ] Component: Typescript SDK
• [ ] Component: IO connector
• [ ] Component: Beam YAML
• [ ] Component: Beam examples
• [ ] Component: Beam playground
• [ ] Component: Beam katas
• [ ] Component: Website
• [ ] Component: Spark Runner
• [ ] Component: Flink Runner
• [ ] Component: Samza Runner
• [ ] Component: Twister2 Runner
• [ ] Component: Hazelcast Jet Runner
• [ ] Component: Google Cloud Dataflow Runner

[riteshghorse]

Actually this work as is for Keras 3. The problem is when the users try to use the deprecated SavedModel format with tensorflow>=2.16 as mentioned here. So actual work here is to raise a meaningful exception if this compatibility contract is broken.