A quick update...
After a fair bit of testing, I think it's safe to say that there is no problem with the Yeast Forge electronics or firmware when it comes to temperature control. The method of attaching and insulating the probe is the critical factor. While the bubble wrap and rubber band recommendation is as simple as can be, I don't think it works well. The bubble wrap is bulky and obscures the probe, which is making it difficult to determine how well the probe is mounted. The large air gaps in the bubble wrap don't really help with insulation. Using bubble wrap is better than having a bare probe, but there must be better alternatives.
I played with a few:
Two sheets of 3mm thick packing foam glued together, with a channel cut out from one sheet for the probe. The probe is mounted vertically, with the cable entering at the top and the probe cylinder wall is parallel to the vessel wall. The tip of the probe is sitting somewhere near the 1/4 to 1/3 distance from the bottom of the vessel. This worked well, but I noticed that after a day or so, the rubber band holding the foam down caused the foam to deform slightly and the ends were lifting a bit. That was easy to fix, I just used two rubber bands to hold both ends of the foam block in place.
An alternative mounting configuration, with a polystyrene mounting block. It holds the probe perpendicularly, such that the tip of the probe is in contact with the vessel wall and most of the probe body is insulated inside the polystyrene. Polystyrene is messy and difficult to work with, so I would probably not recommend it as the best choice of material. Trying to match the curvature of the polystyrene directly to the vessel shape is a never ending exercise in frustration. :angry:
Both mounting methods have greatly improved the accuracy of the target temperature, resulting in an absolute error of less than 0.5C. That's pretty good!