model.register() repackages model_data to session's default bucket when being passed entry_point, source_dir and model_data

[ViktorMalesevic]

Describe the bug
A clear and concise description of what the bug is.

model.register() repackages model_data to sagemaker or pipeline session's default_bucket when being passed entry_point, source_dir and model_data, instead of bucket provided through model_data

To reproduce
A clear, step-by-step set of instructions to reproduce the bug.

Step to register model with custom scripts (entry_point and source_dir) and model_data

import boto3
import sagemaker
from sagemaker.model import Model


boto_session = boto3.session.Session()
region = boto_session.region_name
sagemaker_session = sagemaker.Session()
role = get_execution_role()

model_package_group_name="MY_PACKAGE_GROUP_NAME"

image_uri = sagemaker.image_uris.retrieve(
     framework="pytorch",
     region=region,
     version="1.11.0",
     py_version="py38",
     image_scope="inference",
     instance_type="ml.g4dn.xlarge",
 )

model = Model(
    image_uri=image_uri,
    model_data="s3://MY_BUCKET/somepath/model.tar.gz",
    entry_point="inference.py",
    source_dir="scripts/",
    sagemaker_session=pipeline_session,
    role=role,
)

model.register(
    model_package_group_name=model_package_group_name,
    domain="NATURAL_LANGUAGE_PROCESSING",
    content_types=["text/csv","application/json","application/jsonlines"],
    response_types=["text/csv","application/json","application/jsonlines"],
    inference_instances=["ml.g4dn.xlarge","ml.m5.xlarge","ml.c5.xlarge"],
    transform_instances=["ml.g4dn.xlarge","ml.m5.xlarge","ml.c5.xlarge"],
)

Step to inspect model container definition in SageMaker Model Registry

response = client.list_model_packages(ModelPackageGroupName=model_package_group_name)
response = client.describe_model_package(
    ModelPackageName=response['ModelPackageSummaryList'][0]['ModelPackageArn']
)
print(response)

Expected behavior
A clear and concise description of what you expected to happen.

We should get in response:

...
'InferenceSpecification': {'Containers': [...
'ModelDataUrl': 's3://MY_BUCKET/somepath/model.tar.gz'
....
]

Screenshots or logs
If applicable, add screenshots or logs to help explain your problem.

What we get instead:

...
'InferenceSpecification': {'Containers': [...
'ModelDataUrl': 's3://SAGEMAKER_SESSION_DEFAULT_BUCKET/somepath/model.tar.gz'
....
]

System information
A description of your system. Please provide:

• SageMaker Python SDK version:
  '2.107.0'
• Framework name (eg. PyTorch) or algorithm (eg. KMeans):
  Observed for PyTorch, but issues expected for all Frameworks
• Framework version:
  1.11.0 but likely irrelevant
• Python version:
  py38 but likely irrelevant
• CPU or GPU:
  GPU but likely irrelevant
• Custom Docker image (Y/N):
  N

Additional context
Add any other context about the problem here.