Comment on X-L Meng, ‘multiple-imputation inferences with uncongenial sources of input’