This is practically the reverse situation I encounter, since I measure primarily in daylight hours, often finishing when the temperature is lower than the average while I am measuring. IMHO, both of these situations could be addressed by adopting a SCCF that would "cover" the potential inaccuracies in measurement caused by post-cals in lower temperatures.
It seems to me we would need to accumulate some data on how much measurements are skewed or potentially changed in many of these situations, then graph the results for several measurers. Obviously, different tire types and sizes will produce a range of results. We would then bracket the approximate differences in measurements when post-cal temps are lower than pre-cals.
If, for instance, a post-cal measurement is say, 15 counts higher - 18,000, for example, than a pre-cal average of 17,985, we are contemplating an inaccuracy of less than 4.5 feet per mile. If we had used a SCCF multiple of 1.002 instead of the established 1.001, we would have measured a course needing no adjustment.