FEATURE: Allow the user to reset their benchmark.

[KatieProchilo]

Let's say that first a user creates a baseline model using the sample test data we provide. Since it is the first run of the workflow, this will become the baseline that future models must improve upon.

Now let's say the user finishes the walk through and wants to test the baseline model using real data from their project. This would not become the baseline that future models must improve upon, which doesn't make sense and presents more work for users. They'd have to go into Azure and properly delete the benchmark file.

Instead we can have the user push a tag beginning with RESET. This tag would test the latest baseline model against the test data, and it would become the new benchmark WER to beat in the future.

Now that I've done the baseline changes, this is a super easy add. Initializing an env var and then adding a conditional.