Use equivalent sample size instead of a fixed alpha to define priors

[bmihaljevic]

A fixed alpha limits to K2-type priors, while equivalent sample size allows for score-equivalence and thus seems more natural.