Fix e2e-ml-workflow notebook: pin MLflow version and add NCD deployment dependencies

[lavakumarrepala]

Description

Checklist

• I have read the contribution guidelines.
  • General
  • SDK
  • CLI
• I have coordinated with the docs team (mldocs@microsoft.com) if this PR deletes files or changes any file names or file extensions.
• Pull request includes test coverage for the included changes.
• This notebook or file is added to the CODEOWNERS file, pointing to the author or the author's team.

[lavakumarrepala]

Root Cause

The notebook's conda.yaml uses unpinned mlflow, which installs the latest version. MLflow
2.17+ introduced a logged-models API (/api/2.0/mlflow/logged-models) that Azure ML's tracking server does not
support, causing a 404 during mlflow.sklearn.log_model().
After Azure/azureml-assets#4579 , the platform's no-code deployment (NCD) scoring script
(mlflow_core_script.py) changed its import from mlmonitoring to azureml.ai.monitoring. The notebook's MLflow
model does not declare these packages as dependencies, so the auto-generated NCD inference environment is
missing them and the container crashes on startup.
Fix

Pin mlflow==2.16.2 in the training environment conda.yaml (latest version supported by Azure ML)
Add extra_pip_requirements=["azureml-ai-monitoring", "azureml-inference-server-http"] to
mlflow.sklearn.log_model() so the registered model's dependency metadata includes the packages required by the
NCD scoring environment