My changes to run this with AzD Pipelines

[Damovisa]

YAML CI which will re-build the AML pipeline and re-run it, plus changes to support it. The pipeline in Azure DevOps will need three variables set:

• computetarget
• datastorename
• datastorepath

[Damovisa]

Things I intend to add:

• Restructuring of the code files to the CI only runs if relevant files are modified
• Storage of the inference code in an Azure Artifacts Universal Package, plus a reference to that stored alongside the model when created. This will allow the correct inference code to be used with the model (rather than pulling it from GitHub which may be mismatched)

[sethjuarez]

I cleaned up everything and added some innovations on a separate branch!