feat: Support uploading local models

[Ark-kun]

Added framework-specific model uploaders for XGBoost, Scikit-learn and Tensorflow

This PR adds framework-specific upload functions to the Model class:

• upload_scikit_learn_model_file
• upload_xgboost_model_file
• upload_tensorflow_saved_model

Here is what these functions do for the user:

• Validate model file
• Validate framework version or choose default one
• Construct model display_name
• Create directory for the model (every uploaded Vertex model must be a directory)
• Rename the model file (Vertex AI only supports a small strict list of file names)
• Create staging GCS bucket if needed
• Upload the model directory to a unique location in the staging GCS bucket
• Construct serving container URI based on framework, version, CPU vs. GPU, region etc
• Upload the model

This PR also augments the Model.upload function such that when artifact_uri points to a local model file or directory, the model is automatically staged in Google Cloud Storage as required by the Model.upload API. The Model.upload function now also checks the model files for correctness.

This change does not add any new dependencies.

Example usage:

model = aiplatform.Model.upload_scikit_learn_model_file("iris.sklearn_model.joblib")

model = aiplatform.Model.upload_xgboost_model_file("iris.xgboost_model.bin")

model = aiplatform.Model.upload_tensorflow_saved_model("iris.tensorflow_model.SavedModel")

Prediction example:

model = aiplatform.Model.upload_scikit_learn_model_file("iris.sklearn_model.pkl")
endpoint = model.deploy(machine_type="n1-standard-2")
endpoint.predict(instances=[[1, 2, 3, 4]])

Output:

INFO:google.cloud.aiplatform.models:Using the 0.24 version of Scikit-learn.
INFO:google.cloud.aiplatform.utils:Uploading "/tmp/tmpagz_4y34" to "gs://avolkov-31337-staging/vertex_ai_auto_staging/2021-10-18-19:10:36.387"
INFO:google.cloud.aiplatform.models:Creating Model
INFO:google.cloud.aiplatform.models:Create Model backing LRO: projects/140626129697/locations/us-central1/models/9131602398354079744/operations/3090562396373123072
INFO:google.cloud.aiplatform.models:Model created. Resource name: projects/140626129697/locations/us-central1/models/9131602398354079744
INFO:google.cloud.aiplatform.models:To use this Model in another session:
INFO:google.cloud.aiplatform.models:model = aiplatform.Model('projects/140626129697/locations/us-central1/models/9131602398354079744')
INFO:google.cloud.aiplatform.models:Creating Endpoint
INFO:google.cloud.aiplatform.models:Create Endpoint backing LRO: projects/140626129697/locations/us-central1/endpoints/6185702884286398464/operations/802733785668911104
INFO:google.cloud.aiplatform.models:Endpoint created. Resource name: projects/140626129697/locations/us-central1/endpoints/6185702884286398464
INFO:google.cloud.aiplatform.models:To use this Endpoint in another session:
INFO:google.cloud.aiplatform.models:endpoint = aiplatform.Endpoint('projects/140626129697/locations/us-central1/endpoints/6185702884286398464')
INFO:google.cloud.aiplatform.models:Deploying model to Endpoint : projects/140626129697/locations/us-central1/endpoints/6185702884286398464
INFO:google.cloud.aiplatform.models:Deploy Endpoint model backing LRO: projects/140626129697/locations/us-central1/endpoints/6185702884286398464/operations/8087869132894109696
INFO:google.cloud.aiplatform.models:Endpoint model deployed. Resource name: projects/140626129697/locations/us-central1/endpoints/6185702884286398464
Prediction(predictions=[2.0], deployed_model_id='2568679064810291200', explanations=None)

Thank you for opening a Pull Request! Before submitting your PR, there are a few things you can do to make sure it goes smoothly:

• Make sure to open an issue as a bug/issue before writing your code! That way we can discuss the change, evaluate designs, and agree on the general idea
• Ensure the tests and linter pass
• Code coverage does not decrease (if any source code was changed)
• Appropriate docs were updated (if necessary)

Fixes #<issue_number_goes_here> 🦕

[sasha-gitg]

Please add unit tests.

Test examples with mock Model service: https://github.com/googleapis/python-aiplatform/blob/main/tests/unit/aiplatform/test_models.py#L500

Testing with mock GCS: https://github.com/googleapis/python-aiplatform/blob/main/tests/unit/aiplatform/test_training_jobs.py#L311

Would also prefer to have one integration test as these methods use both GCS and Vertex. Here's an example: https://github.com/googleapis/python-aiplatform/blob/main/tests/system/aiplatform/test_e2e_tabular.py#L50

Let me know if you have any questions.

Thanks for this PR, Alexey!

[Ark-kun]

Please add unit tests.

Test examples with mock Model service: https://github.com/googleapis/python-aiplatform/blob/main/tests/unit/aiplatform/test_models.py#L500

Testing with mock GCS: https://github.com/googleapis/python-aiplatform/blob/main/tests/unit/aiplatform/test_training_jobs.py#L311

Thank you for the great pointers for the tests.

I've added unit tests for the new Model.upload_* methods and also for the expanded staging capabilities of the Model.upload method.

Please take a look.