Simpler manual retraining workflow for non-console applications

In the current version of Model Builder (16.6.3.2136603 at the time this issue was reported), after the model is trained and the training pipeline is generated, it's difficult to **manually** retrain a model in anything other than a console application.

With MBConfig, why might I want to manually retrain?

1. I want to manually tweak the pipeline
2. Change the version of ML .NET used.

For example, if I "Add Machine Learning" to an ASP.NET application or Xamarin application, I can't easily call the `RetrainPipeline` method. While technically I can call that as part of the initialization process, I'll have to comment out that code or delete it once I'm done because the purpose of my application is to use the model, not retrain it. Therefore, there are 2 ways to go about retraining a model manually:

1. In the "Consume" step, I choose to create the console app project just to be able to retrain.
2. Create a standalone console app just for retraining. The issue with this is, I have to manually copy and paste the MBConfig, code-behind, and model to the console app because I can't reference web / xamarin apps inside a console app because they're not class libraries. I also have to manually configure the *csproj* file. \

I don't have a good solution for how to make the experience better, but it would be good to review the current experience to try to find simpler alternatives.