Energy intake in short bowel syndrome: assessment by 24-h dietary recalls compared with the doubly labelled water method