Fix bad installation of tensorflow by pinning protobuf for RAI blbooksgenre notebook

[imatiach-msft]

Description

Fix bad installation of tensorflow by pinning protobuf for RAI blbooksgenre notebook

Still seeing errors in blbooksgenre notebook like the following, but towards the end of the notebook:

TypeError: Descriptors cannot be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
 1. Downgrade the protobuf package to 3.20.x or lower.
 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

This seems to be due to a bad installation of tensorflow in the notebook environment - running the notebook in AzureML compute instance seems to work fine.

When looking at the python installation, it looks like originally "protobuf-3.19.6" is installed with tensorflow, but then the "google-api-core<3.0.0,>=1.0.0" package which is a dependency of azure-ai-ml upgrades protobuf to "protobuf-5.27.3", which is not compatible with tensorflow and generates the error message in one of the imports.
The temporary fix for now is to pin protobuf to a valid version.

Checklist

• I have read the contribution guidelines.
  • General
  • SDK
  • CLI
• I have coordinated with the docs team ([email protected]) if this PR deletes files or changes any file names or file extensions.
• Pull request includes test coverage for the included changes.
• This notebook or file is added to the CODEOWNERS file, pointing to the author or the author's team.

[imatiach-msft]

closing as this seems to cause the azure-ai-ml package to fail instead - instead I will send a fix in the ml-wrappers package for this issue: microsoft/ml-wrappers#176