It works? Maybe? Outside of strange reading spikes during watering (the voltage jumps really high, more than it does when in dry air), it seems to be giving me pretty consistent/meaningful readings. I did notice a lot of variation early on, but I realized this matched with temperature changes quite closely, so after gathering a little data and doing a curve fit in Excel, I was able to filter this out for the most part.
I ultimately adjusted my control scheme to only check the moisture level once a day. I started with it doing it super-early in the morning, but it happens that the faucet is RIGHT next to our bedroom and the flow is large enough that it's quite loud. The water-hammer from it shutting off is even louder, so I changed this to check in the evening, a bit before we go to bed. No more getting woken up by my sentient house.
I also put together an additional sensor and found the readings to be in the general ballpark as the original sensor. Still required calibrating to get them essentially matched up (I just left them both in dry air for a few minutes and then put them in cups of water to get both extremes). I also noticed that the new sensor has much more stable readings and DOESN'T have the strange voltage jumps during watering. Not sure if I did a poor job sealing up the wiring on the first one, or if the few extra months of use is the culprit. One other random thing I noticed; when I plugged in the additional sensor the reading for the other sensor changed as well (can't remember if it was up or down). Not really sure why this is, unless they have enough power draw to lower the actual voltage of the whole system with both operating? Whatever, it works.
I've since moved the 'new' sensor to another part of our garden and set it up to check the moisture level every evening and send my wife a notification that the plants would like some water if it's below a certain value.
I also used this project to solve another 'problem'. We have a small fountain in our yard that regularly requires having additional water added to it. I had already added a 'smart' plug so I could automatically turn it on/off (tied to motion from a security camera) so it would only run when we're in the backyard and shut it off and send myself a notification when the power draw dropped suddenly (the pump power draw is related to flow; low/no water equals less power draw). By adding a second valve to this project and running a hose to the back of the fountain, I've now removed myself from the equation and the fountain is able to fill itself when it gets low (and still shuts off/sends me a notification if a filling cycle doesn't actually fill it up).
Pretty satisfied with the outcome of this project, although time will tell how durable the various components are. I plan to take the whole thing inside over the winter to avoid unnecessary wear and tear. If the sensors can last a season before needing to be replaced, I'd be more than happy.