New and old puzzles in the morphological conditioning of coronal stop deletion