Simple experiment:
Take an empty refrigerator, cool it down for, say 24 hours. Watch the temp stabilize. For bonus points, put a recorder on the temp probe and watch it go up and down (the deadband that DamonHD is talking about).
Now, turn the power off for 5minutes, 10 minutes, 30 minutes, several hours.
You will notice that at short time frames there will be very little temperature excursion, basically within the deadband. Certainly, longer times of no power are going to affect the temperature but no one is talking about hours of power delay - more like minutes.
Even small, 'dorm sized' refrigerators that we use for storing vaccines (and likewise have to be kept within a +/- 3C temp band and monitored to prove it) can handle 30 minute cycles (if you don't open the door). I would imagine a larger fridge would hold for at least 15 minutes even with occasional door opening. But you could certainly check it out, look at potential power savings and decide if the system did what you want.
FWIW a typical home refrigerator will hold temps within the safety range of most foods for AT LEAST 4 hours, a deep freezer for 24 to 48 hours.