...
The problem with self-heating is that we are increasingly sensitive to local windspeed rather than temperature. We definitely don't want to exceed one volt across one thermistor, or 2 volts across both of them.
Paper on thermal time constant measurements in still and moving air: [15200426 - Journal of Atmospheric and Oceanic Technology] Time Constant Estimates for Radiosonde Temperature Sensors.pdf
Current stage: taking long series of data (eg a few days) to a local db to get a sense of noise, frequency characteristics