No star is good news: A unified look at rerandomization based on p-values from covariate balance tests