Change "sd" for sar and fmi regressions to multipliers

[DanOvando]

Instead of overwriting the SD from the regression, pass a multiplier, so that when its 1 you use the regression, when it's smaller you do less than the PP sd.