I see three problems with this:
1) The ping pong ball, or any buoy could hit the edges of a tube and cause an error in the reading
2) There will be some residual CO2 gas that will form inside of the tube if it is closed which could also lead to some error
3) Problems with dissolved CO2 or CO2 bubbling out of solution creating a degree of error.
I think the third (and possibly second) problems would be negligible, or could easily be compensated for, however. it is similar to my idea of using a buoy (such as a ping pong ball or hydrometer) that is submerged and held down by a force transducer.
Also the force of pingpong ball is quite strong compared to a traditional hydrometer, so the adhesion to the pipe wall is probably irrellevant?
You could also make a pipe of stainlesssteel net, to reduce adhesion and ensure not reading a local value inside the pipe,...