Is it sensible to add a "both" option to the Page-Hinkley direction argument?

[macandleighs]

Right now, you'd have to run P-H twice on the same data to check both directions. Would thresholds change?