The form of argument "maybe X invalidates theory Y, so we shouldn't trust prediction of Y" isn't terribly useful since it could be applied to any prediction of science; "Maybe quantum gravity so let's not worry about conservation of energy." "Maybe comoving frame, so maybe we can build antigravity devices", etc. Of course, it could be true. Most anything could be true. It's just not a useful basis for adjusting one's priors.
Actually it is. There is no good reason scientists should all follow the same course of action. In fact, per your own appeal to statistical reasoning, such reasoning is only meaningful if you have an actualization of the samples. If no scientists ever try the unlikely scenarios then you will always reach local optimum.
What is at stake is "Some scientists should consider FTL travel possible, and give it a go." for such a statement, the argument definitely holds water.
