Variable selection for clustering with Gaussian mixture models: state of the art