There is no 100% reliable way.
That's why the real world has difficult challenges. You have to decide what the best way out of several options are and none of them are 100% reliable.
Why would you want to ban a dog that looks like an XL bully but has none of the interbreeding or genetic issues than an XL bully has?
It would be no more likely to snap and attack someone than any other dog.
Whereas in your example, a dog with the interbreeding and genetic propensity to launch unprovoked attacks on humans would be okay because it didn't look quite like an XL bully?
That is madness to me.
I think we need to understand that the root cause of the issues with this breed are genetics.
Yes, these are exacerbated by some of the idiots that own them but a number of the fatalities have been family pets with no prior history of aggression.
They are very, very badly interbred and from lines with known human aggression.
The problem with them isn't that they're big dogs with big teeth. Lots of breeds including my own dog are big dogs with big teeth. It makes no odds because they don't attack.
The problem is their genetic breed propensity to snap and launch unprovoked attacks on humans (at which point the big dogs with big teeth thing does matter).
That's why using DNA is the best option to identify them, not through looks.
(And I do know a fair amount about genetics but I agree that none of us will be making the decision so we'll wait and see...)