Having something like a variable fan can be a good idea, but it can also be tricky. Such a board does not have a single temperature, but many different heat source on the board and thus hotter and colder areas. If you increase the air flow over the board one would not just lower the temperature of the hole board, but have a usually larger effect on the hot parts, thus also changing gradients. This is likely more of a headache. However it can work if the critical sections are inside metal boxes so that you only have a temperature for each box - the thermal design would than still need to couple those boxes or make them move the same way. A fully shielded box around some parts may be a good idea anyway as there is more and more RF around from phones and WLAN.
For the auto-cal part, there might be more granularity. Just for doing voltage readings you don't need the adjustment of the Ohms and amps. Especially some of the volts part only need a rather short adjustment, that does not take long. Other parts, like AC amps might take a rather long internal adjustment, maybe no repeating everything just because of the temperature change. Correcting for temperature drift is already done in some meters (e.g. the DMM7510). However this does not really work with humidity, as this usually is a delayed effect, like the average RH over the last 1 days to 10 days depending on the part and temperature. Also sensors are way more complicated.