Oh, he has a pair of straight-as-an-arrow busbars on the rightmost-terminals, 3 segments each.
Then the next row over uses a ‘snake’ design where a total of 7 busbars are snaked between two columns of cells (twice, each for 1/2 of the rows).
Instead of that, he could have again used 3 busbars to ‘straight as an arrow’ tie the first column of terminals together with 4x1 busbars used to jump between column 1 and column 2 terminals (forming a ‘comb’).
I’s still 7 busbars total but the worst-case resistance to the farthest cell is only 25% worse (5 busbars in series versus 4 busbars in series) instead of 75% worse (7 busbars in series versus 4 busbars in series).
If the maximum number of busbars that can be stacked on a single terminal is limited to 2, the snaked design is the only option. But if a stack of three busbars on a single terminal is easy, the comb design provides less variation in cell-to-cell resistance...
Retreading my post I see that my coffee has not yet kicked in.
Should have been 67% worse (5 versus 3) and 133% worse (7 versus 3)...