I may be wrong, and it could be trainer specific, but along the lines of power meters, I fully expect that trainer makers are tracking and adjusting power reporting based on trainer temperature.
The mere fact that you see temp in the calibration is an indication that they likely track and adjust with it as a reference.
Check with each manufacturer, but the common recommendation is to do a “hot” calibration. That means running the calibration after 10-15 minutes of use (not often defined for resistance level), to heat all the bearings, running gear, and electronics to their “operating temp range”.
When you do this, it takes the spindown time and temp into whatever magic calculation happens behind the scene.
All this gets to the point, that as long as you are using the trainer withing the specified temperature range, I feel it is appropriate the trainer to report “good data”, within it’s stated power tolerance range. I would like to assume that the makers have done appropriate testing at the limits.
I don’t know of anyone who has actually pushed any trainer to the ends (high or low temps) and reported comparisons to actual power meters. So we are at their mercy, unless we have power meters for testing.