Evaluating the impact of keeping indoor dining closed on COVID-19 rates among large US cities: a quasi-experimental design