Mora3 temps might be extremely dependent on ambient temps only because such monster rads are overkill for most common cooled pc hardware, so with very low air-to-water deltas such rads bring, ambient fluctuation might differ relatively much higher. Just like with TIM tests on relatively low heat output cpu-s at stock clocks/voltage - if difference between them is less then one degree, slight ambient change might mess all the results.
At the end imho best way would be _controllable_ flat ambient temps, with eg. some kind of conditioner in room, to worry less about temp fluctuation. Or artificially introduce some insane heatdump in test loop, eg. several water heaters or alike, so that delta is brought up high even with such insane rads