Going back to the lambda probe and monitoring the wave form would give an indication as to what the mixture is doing. The waveform should dither between rich and lean, 0.8v signifying a rich mix and 0.2v a lean mix (the ideal mixture ratio of 14.7:1 occurs at 0.45v). If the voltage reverses (cross-count) less than once a second at 1500rpm, the probe is probably too sluggish for correct operation and replacement is the best cure. Sometimes an extended period of town driving will cause this, and hard-ish use of the engine will fix it, even if only temporarily.
The reference air input to the sensor is actually down the wires to the sensor from the plug at the top. Any cleaner fluid getting into the exposed wires at the plug will degrade the sensor with time, either by blocking the air path between the wire strands down to the probe, or by actual contamination of the reference air side of the probe by cleaner fluid. Incidently, probes that have had the wires soldered suffer because the solder flux blocks the breathing path, they may work for just a few months, but then go sluggish.
I know you've changed the probe, but the above comments may be applicable if cleaner fluid got anywhere near the plug.
There must be a cheaper, engineered, solution to REPLACEMENT of the throttle-body!