Update Azure ML Dockerfile based on reccomendations

[Chuxel]

Now that the Azure ML extension works, I spent a bit of time with it and noticed the Dockerfile was missing some things that made the experience have some hiccups. This PR addresses these by:

1. Set the Python path in user settings and forcing the Docker extension to be installed as a workspace extension - I missed this one when we did the pass for python. This also causes an alert to appear until you pick one.
2. Using Anaconda as a base instead of pure Python - Required to run locally for some scenarios
3. Installing the Azure ML SDK and recommended extras - this avoids errors when you start

It then filters out warnings.