Plotting Raster using values from RAT

Hi

Is there any way  to colour a raster by values in a raster attribute table
that  do not correspond 1:1 with the levels() /ID value in the raster
attribute table.

The example in the rasterVis documentation
https://oscarperpinan.github.io/rastervis/#factor has the categories
aligned with the levels.

Here is what I would like to achieve
r<-raster(matrix(sample(1:9,100,replace=T),10,10))
plot(r)
levels(r)
r<-ratify(r)

df<-data.frame(ID=1:9,
               Grouping1=sample(LETTERS[1:3],9,replace=T),
               NewValue = rnorm(9),
               Grouping2 = sample(LETTERS[4:8],9,replace=T))

rat<-merge(levels(r)[[1]],df)
print(rat)
levels(r)<-rat
plot(r)
rasterVis::levelplot(r,att="Grouping1")
rasterVis::levelplot(r,att="Grouping2")
rasterVis::levelplot(r,att="NewValue")

# in all three cases the colouring of the raster remains the same (
according to the levels() or ID).
# and the legend  only adds a label to each of those levels
#I would like to be able to  colour the raster according to the unique
values in the "groupings"  and the legend to have one row for each unique
value - eg Grouping 1 legend would have 3 colours  corresponding to A B C
or on a colour scale for the numeric "NewValue", without reclassifying the
raster.