Supercedure arises from pheromone responses (or lack there of more correctly).
So for supercedure to occur, there must be a significant lack of the queens pheromones (or of her brood) or serious notable injury. The lack of pheromone isn't as notable when the bees cluster (so supercedure must occur, if it is, before temps get cold enough to need to cluster). Keep in mind, temps could be low enough that it may not be practical for bees to cover supercedure queen cells (if the cell fails, the existing queen remains). I don't have scientific proof, but I believe that the colony expectations of the queen to lay plentiful declines as the season changes. This may account for lower fall supercedure rates.
[Does there come a time when the bees know that it is too late...]
So your answer is yes, and no.
Do they know, probably.
Will they still try, you bet, its survival we are talking about here.
The fear should be that a supercedure queen emerges, kills the old queen and is not properly mated or killed in a mating flight.
[If it's too late, they probably won't have any eggs to make a new queen from anyway.]
Almost all my hives have some degree of egglaying nearly year round.
I had some early russians that were an exception during dearths.
If despirate enough, the bees will try to rear from any single egg (even a male egg in the case of laying workers).