All of these things have a pattern in common where the majority of the improvement can be had reasonably cheaply and capturing the last small improvement has huge costs that are usually much more costly than the benefit of eliminating the last bit and the population of potential events is huge. Road deaths are an illustrative example. Adding speed limits, traffic lights, crosswalks, moderate enforcement, etc. all dropped road deaths precipitously in the USA. If the USA changes its model from cars first to people first there are still some moderately costly infrastructure changes they could make to get their numbers down to Netherlands levels, which are still not zero. Beyond that you are left with hugely costly heavy enforcement among steadily more dystopian and invasive methods that will likely improve the number but not get it all the way to zero. It is really unlikley that these last improvements cost less than the benefit they incur between the direct cost of implementation and the negative effects on the total population of driving events.
On top of all this all those large costs you are incurring for very small gains come at the expense of being able to spend those limited resources on other things that likely have better cost to improvement ratios. You are foregoing a larger decrease in something else bad for society for small decreases in your target. In the road deaths example above that final small drop in road deaths might be coming at the cost of large drops in crime if you deployed those police there instead of on road deaths, as an example.
Maybe your writing was not clear enough, m'kay?
And now you play word games with my intent. I did not advocate for extreme measures -- simply to have clear targets of and rules of engagement; to identify workable solutions.
You're manufacturing slippery slope reductions of a concept and therefore dismissing the concept.
What should the goal of police departments of pet dogs killed by cops? Should they aim for 10,000? A million? Just a handful? A logical answer is "we want to avoid killing any pet dogs in the course of service" (also known as zero).
Now they could take absurd actions to avoid that (leaving the scene whenever a dog is present), or they could add that to standardized training so that they're better prepared to deal with that situation.
So without providing real-world examples of you concerns you are just offering up florid conjecture.
What should the goal of the police department regarding killing innocent people? Should they aim for a dozen per quarter? Per-capita weighted?
Incremental changes have been hard because it gets down to the police policing themselves, and that has shown to be a failure.
Again, one of the "simplest" reforms would be to have other professionals deal with mental health crises.
A bigger reform with much more value would be ending the War on Drugs but that's way higher up the food chain. It is germane though, in that the whole point of said war was to give police more opportunities to oppress "others" (i.e., "minorities" and "hippies").