adding s3 bucket parsing for model paths

[jasonrusselwang]

When using import-model to import from an exported open source project to Databricks, I was receiving the following error:

mlflow.exceptions.MlflowException: Model version creation failed for model name: help-mlops_reference_project-normal_tickets version: 2 with status: FAILED_REGISTRATION and message: Failed registration. The given source path `dbfs:/databricks/mlflow-tracking/2485056402562980/6ddc7ac453ea44abb2f740628fe3b423/artifacts/45/007930631053402f957be01de073fc3e/artifacts/model` does not exist.

This is due to the source field in model.json being an S3 bucket that contains "artifacts" (i.e. s3://test-analytics-us-east-1-mlflow-artifacts) which results in the pattern "artifacts" occurring twice, resulting in:
dbfs:/databricks/mlflow-tracking/2485056402562980/6ddc7ac453ea44abb2f740628fe3b423/artifacts/45/007930631053402f957be01de073fc3e/artifacts/model
instead of:
dbfs:/databricks/mlflow-tracking/2485056402562980/6ddc7ac453ea44abb2f740628fe3b423/artifacts/model.

I've added a check for source S3 URLs and parses the path so that the bucket name will no longer interfere with the creation of the model_path variable.

[jasonrusselwang]

@mac-macoy
@safwanislam