[development] Remove TF Serving & Use Legacy Keras

[othakkar]

Existing Sample Changes

Description

Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context. List any dependencies that are required for this change.

• Remove TF Serving from the jupyter notebook and add it to the README for user's reference.
• Replace tf.keras with tf_keras to use the legacy Keras (2.x) version.
• Choose precision for inference based on the underlying architecture.
• Update requirements.txt to include neural_compressor (used to load the model for inference) and py-cpuinfo.

Fixes Issue#

• CI was failing as the port for TF Serving was not exposed to the public.

External Dependencies

List any external dependencies created as a result of this change.

Type of change

Please delete options that are not relevant. Add a 'X' to the one that is applicable.

• [X] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] Implement fixes for ONSAM Jiras

How Has This Been Tested?

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

• [X] Command Line
• [ ] oneapi-cli
• [ ] Visual Studio
• [ ] Eclipse IDE
• [ ] VSCode
• [ ] When compiling the compliler flag "-Wall -Wformat-security -Werror=format-security" was used