it would be helpful to know what locations were searched when butler can't get a location for a dataset while performing Butler.get.

IE if the Butler is creating a ButlerLocation to read from, and a repository is able to complete the template (from the passed-in dataId & registry lookups), it then searches to see if that file exists in that repository. If not, it continues searching. In the "if not" case, a debug trace should be logged, with the storage root and the completed template, stating that the location was searched, but no file was there.

It's not immediately obvious to me where the logic to emit this trace goes, but I think it's doable.