set up an mlflow 3 deployment job on databricks to gate model version promotion

Steps

1. Create model versions in a Unity Catalog-registered model as the trigger source for the deployment job
2. Build a deployment job with the required evaluation, approval, and deployment steps, each implemented as a notebook task
3. Set the two required job parameters, model_name and model_version, so the job knows which version it is acting on
4. Have the evaluation step call mlflow.evaluate to produce validation metrics on the new model version before any approval gate
5. Configure the job to run under a service principal with minimal permissions and set the max concurrent run limit to 1 (the recommended default) to avoid deployment races
6. Register a new model version to Unity Catalog and confirm the deployment job fires automatically, with its run status visible on the model and model version pages