Automatically trained peak scoring models should be renamed to "Auto" or "Auto (mProphet)", rather than the document name.
Also, we should make sure that automatic models are only available to the document it was trained from.