John, my thoughts are... it depends
Seriously though, it will depend...what I would be looking for is consistency of the dark frames. How effective the cooler is at regulating the temperature of the sensor will very much impact what's in the dark frames. If there is too much variance, a larger set of dark frames won't necessarily be better than a smaller set, but introduces its own challenge as to which set you apply
Btw, my impression (and it is just that) of lucky imaging was that it used very similar techniques to planetary imaging...thousands of very short (sub-1 second) subs with some capturing good detail in the fleeting seeing.