* add tests that highlight known issues in the destroy mechanism * separate refresh during destroy plans * use the refresh outputs during destroy plans * copywrite headers