Controlling the quality of metabolomics data: new strategies to get the best out of the QC sample