Hmmm...
Seems like totally normal behavior to me. (At least, mine does the same...)
The direct (with BNC to croc-clip cable) connected calibration signal shows a perfect rectangular 3VPP wave.
All four probes have the same drift after calibration. (Not immediatly. After some time.)
Possible explanation:
The wiper of the calibration trimpot in the probe is made of a cheap metal (maybe aluminium).
This metal is oxidizing when in contact with air. It builts up a layer of oxid on the surface.
This layer prevents the inner part of the metal from further oxidising.
If the wiper is moving while a calibration, then some oxid is scratched of the surface.
After calibration it begins to oxidising again, until the surface is complete covered.
Possible remedy:
Apply some contact-spray or much more simple, a little amount of oil to the trimpot(s).
This covers the metal surface, avoiding the contact between metal and air and thereby the oxidation.
Then there is no more ongoing oxidation, hence the electrical parameters of the trimpot should be stable.