On the Watson L2-theory for index transforms