I agree, there's either a problem with the data, or the tire compound has changed over time. It would help if BRR tested multiple samples of each tire. Also, the 2 older tests were run with tubes and then the tubeless Crr was later calculated when they updated test protocols.
From the BRR...