On the detection of both carbonyl and hydroxyl oxygens in amino acid derivatives: a 17O NMR reinvestigation