George Booth, Steve Dixon and myself all reported independently back in the late/mid 1990's that the main issue was the overflow.

Gas loss should be the same otherwise if the sump/dry section of the wet/dry is sealed.

Often they have vents to add air into these sections.
Tape those up.

Bacteria on the biomedia produces CO2 and uses O2.
Sio the in/out result => no net gain or lost of CO2 really and if anything, a sealed dry section would give you a bit more CO2..........

Same with a canister.

But the overflow should have the level raised up to about 2" below the surface, not 6-12-18 inches etc.

This will also reduce noise and evaporation.
So add the media back, tape up any leaks etc in the wet/dry section, feed the CO2 reactor output into the suction side of the return pump.

Be careful after adding these changes, watch the fish so you do not gas them.

Regards,
Tom Barr