So the magnetometer doesn't affect attitude, so that wouldn't be it. If the magnetometer were incorrectly installed, only heading suffers.
Also, a Trutrak ADI will return to level if you hold the aircraft in a constant bank but are not turning, so it's not a good reference for this comparison. It's only behaves like an pseudo-attitude indicator when you actually sustaining a turn with yaw rate (it reacts to bank angle changes too, but only for a moment). Try slipping the aircraft (holding the aircraft banked while using opposite rudder) to create a sustained bank angle without an actual turn to see what I mean: the ADI will come back to level after a few seconds.
The root cause here either that the aircraft is actually slightly banked when in climb attitude (this can be an airframe / airflow effect, possibly), or it's installed slightly tilted in the panel (though if it evens out once you're straight and leve, probably not), or the D100 itself is in need of a recalibration at our factory. In non-accelerated climbing flight, you can take a level and put it across the top of the EFIS to see if there's any "tilt" in the unit's mounting in that flight condition.
I usually set the aircraft level first & hold that with jacks, then do the magnetometer/ADAHRS levelling.