I've done some testing on my own, a couple years ago, and the difference is very small. I even tried different heights to see if i could get a better response, but there wasn't much difference. Smoother in a couple places, but not nearly as drastic as the above results.
The measurements in that graph are most likely near-field measurements, where the round-overs will likely have a larger effect.
In my testing standard testing at 1 meter any differences were quite small, with most changes being +/-2dB happening between 5-5KHz.
When I tested then again in Danny's rig, the differences were even smaller.
The best layout was a tall one in the middle with two round rods.
I probably still have some rods floating around that I can test for more accurate measurements and save the measurements. To share here. Just depends on how long it takes me to find/print more rods our of PLA instead of flexible resin.