No, I don't think I can answer the question
a priori. I think the discrimination test may boil down to a simple experiment though: two identical glasses of water, except one distilled and one brine. Compare the probe readings on the two solutions. Hopefully, within some as yet undefined tolerance, they are the same value. If they measure wildly different, then it's not a good circuit.
The same test could be applied to to pre-made sensors, so if I had those sensors already, I would try it. For sure, I would try it before buying 30 of them.
The fringing approach (above) seems to lend itself more to a DIY solution, as it doesn't require making your own PCB or buying somebody else's PCB. Besides, as you point out, PCB's can have an issue with moisture penetration into the board itself, effectively ruining it, if extra steps aren't taken to thoroughly moisture-proof it. Perhaps Joe's or Chirp's method would work with a couple conductive rings, instead of PCB's.
To your point, most of the people (at least the ones whose comments I've read) who own either the Vegetronix or the Watermark seem to like them. I've already enumerated the most common objections to the watermark. I'm not sure enough people actually own the Vegetronix to have heard whatever negative feedback might exist. It has no forum, and comments are scattered. In addition, Decagon makes moisture sensors that utilize TDR, which is supposedly even more immune to salinity effects than capacitive moisture sensors, but it also costs about $100 a pop. However, having one might be a worthwhile yardstick for calibrating expectations. They sell the scientific community, so they have good documentation. In contrast, Vegetronix claims to work "by magic." That said, it can work by vodoo as far as I'm concerned if it's 1/3 the price, provided that it really works. I'm honestly surprised there aren't better choices, given that this is a near universal problem to be solved. I probably will get a tensiometer, as those are the gold standard, though it measures permitivity, not volumetric water content like the others.
I have doubt that capsense will cope adequately with salinity, if only because it probably would already be in widespread use for that purpose if it were effective.