The coverage processor seems to support some way of using content-based section IDs since recently. This might avoid the re-arrangement of existing tests when new assertions are inserted.