I am wary of facial recognition, and I avoid the use of it. But I'm not convinced by this line of reasoning either, so let me play devil's advocate.
> But again, once authorities have any of this data in their possession, abuse always happens. Literally always.
Well, before something can be abused it must first be available to use. Conversely, once a tool is available to use some may abuse it.
For example, if collecting fingerprints or DNA were completely forbidden then that might prevent abuse of such data (such as false matches). But it would also prevent any beneficial uses as well.
Banning facial recognition prevents not only abuse but also any potential good uses, such as locating victims of abduction or trafficking, and perhaps other uses we cannot foresee.
Killing it in its infancy may be easier than doing so after it takes root, but it also gives society less opportunity to learn what the consequences of the technology may be, intended and unintended, good or ill.
We know it can be abused, especially in the hands of an authoritarian government, but does that mean it cannot be used responsibly? Anything that gives the state power could be turned against the people, as libertarians might warn, but social progress also requires that we learn to work together rather than reject anything which might do us harm.
Perhaps a better argument for an early and complete local ban might be that it allows other regions to be the test subjects. Or that by taking a less compromising stance the anti-facial recognition side gains a stronger bargaining position at the table. But those arguments are not as attractive, maybe.