Consider adding OSC-style conversion for F with df1 > 1

OSC 2015 calculated r as the correlation per predictor - this appears to be an ad-hoc metric, yet it *could* be a useful way to compare effect size shrinkage and similar relationships in factorial designs.

See https://osf.io/z7aux and decide whether to implement as an effect size conversion option.