MMP needed fixes

  1.  If u has two levels, loess goes nuts.  Check and fix.
  2.  mmps copes with missing data, but mmp does not.  DONE
  3.  mmp should see if u is in the data frame, and use it if it is DONE
  4.  I've changed colors to lineColors.  DONE
  5.  Allow a Groups arg that will draw the plot for the specified group
  6.  write mmpControl to control margins in the plot.



# fit the model of interest:
unb<-glm(itch_ep~dog + trtdose,family=poisson)
# get the fitted values and the residuals:
real.answer <- anova(unb)[3,]

# do the bootstrap by resampling Y while fixing the design.
out <- NULL
for (i in 1:999) {
 ystar <- sample(itch_ep, size=length(itch_ep), replace=TRUE)
 out <- rbind(out,anova(update(unb,ystar~.))[3,]) }

