Safe reserves and efficiency
The experiment (as originally described) was clearly a multi-generational one, and the amount of power reserves required for it to survive through a lunar night are not actually all that high. However, 'not all that high' could still quickly become higher then 'safe reserves available for secondary functions' if the units insulation, solar panels, battery reserves or any of a dozen other factors were even slightly less efficient then planned, leaving them no choice but to pull the plug. I still say they deserve major credit for even trying.