Fix tensorflow import in dataset wrapper to handle invalid tensorflow installation due to unsupported protobuf

[imatiach-msft]

This PR is mainly to fix one of our nightly notebook tests that is failing. In the notebook test we are using the EndpointWrapper from ml-wrappers but getting the error for the file https://github.com/Azure/azureml-examples/blob/main/sdk/python/responsible-ai/text/responsibleaidashboard-text-classification-blbooksgenre.ipynb:

2024-08-19T18:56:24.3991935Z     615 def __new__(cls, name, full_name, index, number, type, cpp_type, label,
2024-08-19T18:56:24.3992539Z     616             default_value, message_type, enum_type, containing_type,
2024-08-19T18:56:24.3993063Z     617             is_extension, extension_scope, options=None,
2024-08-19T18:56:24.3993504Z     618             serialized_options=None,
2024-08-19T18:56:24.3993993Z     619             has_default_value=True, containing_oneof=None, json_name=None,
2024-08-19T18:56:24.3994691Z     620             file=None, create_key=None):  # pylint: disable=redefined-builtin
2024-08-19T18:56:24.3995311Z --> 621   _message.Message._CheckCalledFromGeneratedFile()
2024-08-19T18:56:24.3995727Z     622   if is_extension:
2024-08-19T18:56:24.3996143Z     623     return _message.default_pool.FindExtensionByName(full_name)
2024-08-19T18:56:24.3996499Z 
2024-08-19T18:56:24.3996669Z TypeError: Descriptors cannot be created directly.
2024-08-19T18:56:24.3997428Z If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc 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.
2024-08-19T18:56:24.3998317Z If you cannot immediately regenerate your protos, some other possible workarounds are:
2024-08-19T18:56:24.3998940Z  1. Downgrade the protobuf package to 3.20.x or lower.
2024-08-19T18:56:24.3999803Z  2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

After close investigation, it looks like one package, azure-ai-ml, is installing google-api-core which depends on a newer version of protobuf, after an older version of tensorflow==2.9.3 was installed with an older protobuf range. When ml-wrappers tries to import tensorflow for a helper utility, the import fails with TypeError. This PR prevents this error from occurring by ignoring the TypeError and printing a debug message:

>>> from ml_wrappers.model import EndpointWrapperModel
2024-08-21 10:53:19.392103: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
2024-08-21 10:53:19.392270: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Could not import tensorflow in ml-wrappers due to TypeError, required if using a Tensorflow model. Inner exception: 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).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
>>>

Also, as part of this PR the ml-wrappers version was bumped so a release can be done, the is_sequential utility method was updated/fixed to work with latest tensorflow version and some tests were re-enabled.