So while many things in science might be highly unlikely, few should be deemed impossible. After all science is evolving and changing and if we as "scientists" (I am not one but I did stay at a Holiday Inn Express last night) simply write things off as such then we begin to "believe" and our study drifts to dogma rather than openminded science.
Sorry, this has been a pet peeve of mine with science and archeology for a while now.
***Warning, unscientific and unresearched post coming. I am not a scientist!

*** (I just think about stuff a lot)
While I think we would have seen this in nature if it were possible I think this is a great starting point for a thought experiment.
Could it be possible for something like this brackish torch to adapt to freshwater given the right conditions? What would determine success/failure? Is it simply salinity? Is it a combo of salinity and other water chemistry? How would you test for it?
How would I do it if I REALLY wanted to test it out?
I would set up a series of small tanks on racks, say 3 rows of multiple columns. Each tank getting the exact same temp, flow, filtration, feeding, and lighting regime.
Column A would be my control. All saltwater, the way a "normal" tank would be set up. And monitored for everything possible to ensure levels of appropriate parameters stay consistent across the test tanks
Column B Would be the "full change" column where I am changing all water parameters at once.
Column C would be salinity changes only, dosing to maintain ph, alkalinity, calcium, magnesium, etc.
Column D would then be salinity plus one other parameter change.
And so on down the road.
Then I would probably test each row with different "fresh water" types (or keep them all the same for multiple sets of results).
Row 1 being straight RO, unlikely to succeed there would be my guess.
Row 2 might be fresh water with target parameters like one of the rift lakes with ph, alkalinity, mineral content to match.
Row 3 might be fresh water built with different parameters than the above.
Slowly changing over the course of the test period. In this case I think using a clone, fragged heads of the aforementioned torch, would be ideal. That way you have the same organism in each set of water conditions to see if it is plausible, reducing variables.
Granted, from a biological/physiology stand point I am not even sure this is possible with a torch, or anything else. And if I remember right don't most brackish fish/animals start in freshwater as juveniles and transition to salt as adults (or this used to be the belief, the science might have changed on this one), rarely living their whole lives in one or the other?
But this testing methodology probably wouldn't get a lot of hits on youtube...