If people could take the 'dog's don't lie' belief system out of the equation I think people would realise that there is still a great deal of scientific research and work going into understanding just exactly what it is that causes the dogs to react.
Because Shannon was very obviously alive they had to find out what caused the alerts, would they have bothered otherwise?
If her remains had been found or she had not been found at all ... what would that have meant for the investigation and the misdirection of resources into her case.
I agree with everything you have said in your post although I know that dogs do alert in areas where bodies have lain (eg ... soil contaminated by a corpse is used for training purposes and I read that when the containing jar is opened it is possible to discern a scent ... can't find the cite).
Were I on a jury listening to cadaver scent evidence where remains or fragments had not been found ... I would not be a happy bunny.
**Snip
But in the field, VR dogs can sometimes be distracted by “false positives”, such as dead animals, or even mushrooms, explained Lorna. If she can arrive at a greater understanding of the chemistry of odours from human cadavers, then VR dogs can be extra efficient.
https://www.hud.ac.uk/news/2014/august/forensicsresearchtomakecadaverdogsmoreefficient.php
I know these dogs can scent many things that we are unable to even smell and they do a great job in assisting the police in many cases.
However, when they are being tested the examiner knows what has been introduced to the dogs on the test.
No amount of testing can replace a real crime scene.
Nobody knows what exactly the dog is alerting too or why, when it comes to an alert in a real investigation(unless there is a body of evidence).
Alert of blood or other body fluids, may be corroborated at the laboratory, but this is not the case in, (supposed) residual scent.
So as Mr Grime said.......................................................