Similar Experience
During flight testing I had a very sensitive CO detector in the airplane. I found that it would go off when at a high power setting, nose high attitude and low airspeed. I did a very good job of sealing the firewall and found no significant source of air leaks there.
Note the position of the exhaust tube through the bottom cowl relative to the leading edge of the wing. What I believe I was experiencing was that in that condition the airflow around the front of the fuselage changed a little and the exhaust stream was entering the slipstream around the fuselage and entering through the gap between the canopy and the fuselage. I did put some weatherstripping around the base of the canopy and it got a little better, but I was never able to completely alleviate this situation.
The CO detector that I used started to indicate at 8 PPM and I usually saw 12 PPM in climb. It never went above 20 PPM. In cruise it would drop off to 0. I just kept these umbers in the back of my mind and tried to minimize high power low airspeed operations.
I also noticed that the detector would go off intermittently in cruise above about 8000'. I just attributed that to the detector not being specifically designed for aviation and was probably fooled by the thinner air.