Currently CBR calls extractPlanExplainerData, which expects that there are rejected solutions.

This was a useful tassert during development to express an expectation of the MultiPlanStage state, but makes this usage of MultiPlanStage brittle. MultiPlanning is used as effectively "just another ranking strategy" within some CBR "meta" strategies which combine MP and CBR.

But this tassert turns into an implicit contract on such strategies - requiring that it is only invoked when there are multiple candidate solutions, as that is a requirement for there to be rejected solutions.

Note that generally MultiPlanner can handle single solutions, and is specifically used in this way to gather the info required to add solutions to the plan cache, when the plan was selected by CBR.