Evaluating the impact of a national naloxone programme on ambulance attendance at overdose incidents: a controlled time–series analysis