Given the explanation above, this now makes a lot more sense: if the (non-eutectic) solder has cooled gradually, some the higher melting component will freeze first at the coolest location, and things will move around, leaving less of the "low-melting" components in the remaining molten solder. Once it's all solidified, some areas will have more tin and other areas will have more of the other elements. This will induce some weirdness, maybe the tin will want to diffuse to other parts where there is less (is this possible?) or maybe just there will be areas of relatively pure tin, perhaps more susceptible to whisker formation?
On the other hand an evenly-frozen mix of lead and tin (eutectic) wouldn't have these non-uniformities as there would be no driving force for it. One location freezing before the other wouldn't change the composition of the remaining melted solder.
I would also imagine the very quick solidification also prevents larger crystal formations. Crystal growth itself can also lead to separation of the metals and the extra few seconds or more of a half-liquid state solder could represent decades or more worth of crystal growth in ambient temperatures.
On the other hand an evenly-frozen mix of lead and tin (eutectic) wouldn't have these non-uniformities as there would be no driving force for it. One location freezing before the other wouldn't change the composition of the remaining melted solder.