Edit: That is something I dislike about this test, or more the conclusions being drawn from it: the squads at the top are at the top because this test favors them. Cover engagements favor jli. Max vet for all favors falls and jli. The engagement being at long range works against the jaeger command squad and panzerfusiliers.
Even more so the fact that the way these tests were set up (a tested squad facing the previous tested squad) does not produce conclusive results. Or the fact that the incredibly small sample size can potentially favor units that rely more on RNG than others (JLI sniper over Obers LMG for example), or disfavor them. Or the fact that the conclusions drawn do not consider unit cost, timing, role or context. For example Guards do less AI damage for sure, but they have a huge advantage in bringing some light AT as well.
It also seems to use units that are not even remotely comparable (LMG Grens, Jaeger Command Squad, Guards, Panzerfusiliers, etc. all have completely different roles and strengths) to make the top squad(s) seem even better than they really are, although I'm not sure if this was done on purpose.
If one wants to make a truly competetive list, the tests need to be run in a considerable sample size (for the sake of reasonability I'd say 5-10 runs per squad at least, though of course preferably something closer to 100), done in a completely similar testing environment (all tested squads facing the same squad), at different vet levels and in different situations (no/yellow/green cover), comparing only squads that have a very similar role (JLI/Obers/Paras for example) and using something like Time To Kill as a neutral measurement. Then this Time To Kill would have to be brought into perspective by bringing in things like unit cost into comparison.