This combination should account for 30% of possible feature triggering combinations and yet it is much lower than this.
These anomalous results alone strongly suggest to me weighting or grouping and/or the use of algorithms even though there may still be random element.
He already gathered the data, just look at the reels for Moonshine and it's obvious why you will get 2 scatters on the first 2 reels and then nothing.
Both the first 2 reels have 35 stops and 2 scatter bottles, while the last 3 reels have 38,39 and 40 stops with 1 bottle on each reel. Unless I fail at math you have about 22% chance to actually hit a 3rd scatter starting with 2 on the first 2 reels which you will get 1/34 spins.