Estimator local mode do not support dependencies

[idanmoradarthas]

Describe the bug When running the object sagemaker.estimator.Estimator in local mode:

sagemaker_session.config = {'local': {'local_code': True}}

dependencies are not copied to the container.

To reproduce

1. We created HPO step with an estimator in it.
2. We configure the session to be local (see above)
3. We run only the estimator with fit
4. We received an error that it can't find our dependency code

Expected behavior

1. A similar interface such as processing input which allow you to control were in the container the code in being placed
2. Running the estimator in local mode, copied the dependencies to the right folder

System information A description of your system. Please provide:

• SageMaker Python SDK version: 2.126.0
• Framework name (eg. PyTorch) or algorithm (eg. KMeans): sagemaker.tuner.HyperparameterTuner
• Framework version:
• Python version: 3.9.12
• CPU or GPU: CPU
• Custom Docker image (Y/N): Y

Additional context We had to create the following workaround in order to it to work:

# Add repository to container so the code can use its dependencies'
tuning_step.inputs.update({'pipelines': 'file://{}'.format(str(Path(BASE_DIR).parents[2]))})
# Add The above pipelines stream to containers' PYTHONPATH so packages could be found
tuning_step.tuner.estimator.environment = {"PYTHONPATH": str(PurePosixPath("/opt/ml/input/data/pipelines"))}