It's no different than a cansiter filter if it's sealed.

What goes in, must come out.
Same with a canister.

It's sealed as well.
Rather simple concept

So if the dry section is sealed, why would it be any different than a canister?
If you measure the CO2 closely, it's not.
The loss is due to air exchange with an unsealed section, or due to over flow drops.

You can raise the overflow to the tank level and eliminate that as a possibility and then isolate and focus on the wet/dry section to see.

Once you have a good handle on the wet/dry section, then you go back and check the overflow response when you lower it 1", 2", 4" 12", 20" and so on.

You lose very little at 2-3", but it starts to lose CO2 at 4" or more inches depending on flow rates and mixing.

Regards,
Tom Barr