I agree. The starch conversion from crystal malts was not expected or part of the main purpose of this test. It came out as an extra finding and doesn't seems to be significant when comparing to how much fermentable sugars are within crystal malts.
Comparing average attenuation of each of the three malts tested:
10+2R=77%, avg(10&2R)=64.65, increase of 13%
40+2R=69.5%, avg(40&2R)=59.45, increase of 10%
120+2R=67%, avg(120&2R)=59.1), increase of 8%
The higher the lovibond, less startch, lower gain. This seems to make sense to me.
I also agree about the carapils. Need to find the time to do a batch test on this.