Model validation improvements

[mingyu89]

I kept all the example code in model validation. Because the project itself is based on MLflow recipe example.

The PR contains the following changes:

• Extract all parameters from notebook ModelValidation.py to terraform file
• Create validation.py that defines methods that return custom metrics and validation thresholds.
• If fs is enabled : hide env parameter and show targets and model_type
• If fs is disabled : show env parameter and hide targets and model_type
• Update doc
• print message in the notebook about run_mode

There will be followup PR to handle the the parameter that can be directly inferred from MLflow recipe.

---

test

Test 1 : Generate project without feature store, copy terraform files, validation.py and ModelValidation notebook to CUJ project and then run workflow. Test looks good. https://adb-3253686410953403.3.azuredatabricks.net/?o=3253686410953403#job/741125644266775/run/1675013

~Test 2 : Generate project with feature store, copy terraform files, validation.py and ModelValidation notebook to CUJ project and then run workflow.~ The test needs some followup.

---

Make the same change in example repo https://github.com/databricks/mlops-azure-cuj/commit/dd092f62e4a054fcfc21151690456fd4df43d7f6 Manually run the notebook

Run training workflow