I highly doubt there is anything wrong as such. My best guess is that your aggressive forced air cooling is affecting the temperature compensating circuit where the sensing part may not be subjected to the same cooling that the transistors are subjected to. In heavy industry, we have all kinds of instrumentation and control systems that are affected by temperature so there are naturally all kinds of temperature compensation systems in place, and one important point is to make sure if any sensing circuits/devices are involved, they have to be subject to the same ambient temperature as the device/system that they are designed to temperature compensate.
The AVP-A1HDC1 packs tons of components in one box, as such, if subjected to forced air cool, especially a sucking one, it is possible that the sensing/control parts and the parts being temperature compensated are no longer operating per design. The key is to limit the air flow, slower flow should result in more even cooling of all the components inside the box.
I have my laptop running REW monitoring my AV8801 all day, and so far I have seen a maximum variation of about 2 to 2.5 dB when I played around to create extreme conditions, no noticeable difference most of the time. There was one observation though, when I first turned on the system, it was stone cold, yet the output was about 5 dB higher than any other time. That was a weird one as I could not make it happen again, even after running two fans full blast with the unit on standby, until the temperature gets down to room temperature, a little higher because of the standby power. So the lowest temperature recorded on top of the case was 27 deg. C and the highest was 43 deg C. No noticeable change in output so far. I am still monitor it in case something weird may happen again, but no luck so far.
Edit: I found out why I momentary drop 5 dB at one point. For some reason, all of a sudden, REW change the default output from both channel to just left channel, could be just operator (me) error.