We do have a preferred index selection...but it is only used for index writing. Since there is a misalignment of the assumptions of the query strategies (all indices in the same name space contain the adapter data) and the preferred index selection, this seems like a good time to consider making an adjustment here as well. Allow the preferred index selection to filter the set of indices inspected during the strategies analysis.
In performing the analysis, it was noticed that the preferred index selection is not sufficiently persisted in the data store.