Hard numbers? The long-term decline in violence reassessed. Empirical objections and fresh perspectives