I know a few people who do it on a regular basis.
I don't actually know how sensitive the steels we usually use *really* are to temperature fluctuations of the magnitude seen in domestic ovens, whether convection or toaster.
I did some measurements on my first homebuilt HT oven and found the fluctuation at tempering temperature was substantially greater than I'd expected, and much greater than I personally felt was acceptable.
I also used a datalogger with several different-sized thermocouples to try to get a feel for how the radiant heating would affect a blade differently depending on thickness. The overshoot on very small-diameter thermocouples was huge: It suggested that the effect would be minimal near the spine, but near the edge it could be enough to lose several Rockwell points.
As a result I did some fairly geeky things within the control system of my second oven to reduce it: very short output cycle time, ramp/soak control (always slow-ramp-to-temperature), select a fast-response thermocouple and position it for the fastest response practicable. All these are things that need to be done at the design or build stage.
Using thermal mass to damp out the fluctuations seems like an obvious thing to do when you have a completed oven and there is no access to the geekier workings of the control system.
I know one guy who has an Evenheat HT oven, which he uses to temper in sand. He uses a stainless fish kettle for the sand, buries the blades in it and sticks the whole lot in the oven overnight, set to the desired tempering temperature.
Most of the others Austenitize in a forge and temper in the domestic oven. The forges use commercial Venturi mixers and can be set to provide very precise temperature for Austenitizing/soaking, but cannot get down to tempering temperatures.
