Support to override model_data when calling Estimator.create_model

[jonsnowseven]

Describe the feature you'd like
Support to override a model data S3 URI when creating a model:

sagemaker-python-sdk/src/sagemaker/estimator.py

Lines 2419 to 2421 in 34cb1c7

	return Model(
	image_uri or self.training_image_uri(),
	self.model_data,

How would this feature be used? Please describe.
It is useful to call create_model from an estimator, when we are using a tuner job, specially for SageMaker Pipelines.

...
estimator = ...
tuner = ...
step_tuning = ...

model = estimator.create_model(
    model_data=step_tuning.get_top_model_s3_uri(
            top_k=0,
            s3_bucket=default_bucket,
            prefix=prefix,
    ),
    ...
)

instead of

...
estimator = ...
tuner = ...
step_tuning = ...

model_class = <specific estimator model class (e.g., TensorFlowModel)>

model = model_class(
    model_data=step_tuning.get_top_model_s3_uri(
            top_k=0,
            s3_bucket=default_bucket,
            prefix=prefix,
    ),
    ...
)

[akrishna1995]

We have added your feature request to our backlog of feature requests and may consider putting it into future SDK versions. I will go ahead and close the issue now, please let me know if you have any more feedback.