term is used somewhat informally to the general blob of "problematic users"

I can see how this would be more useful from the perspective of the person doing the banning, but I don't see why it would be useful from the perspective of the person who is attempting to avoid being banned. Flexible for one purpose, too vague for the other.

Would you be able to properly formalize the kind of behavior you don't want on a website you run, without being too broad or too narrow?

Somebody has probably already done so. Not perfectly, of course. But they've probably already done so. There might even be a description of undesired behavior in an open source context, either as part of a free legal terms of service agreement, or as part of a piece of open source software. It is quite possible that a good free description has already been written and just needs editing. It's also possible to do better than be flexible/vague and provide a list of behaviors (such as the one you created above) that briefly describes the main concerns, without it being perfect, and simply aim to make an improvement on flexible/vague.

if the policies hit a broader target and also discourage non-trolling obnoxious cranks and idiots

The problem is that people with idiotic ideas do not know they are being idiotic, and I think that although some cranks do know that they're wrong and are content trying to scam people, other cranks are just as clueless as their customers, and have no idea that what they're selling is a ripoff. For instance: I'm not religious, but do I consider a priest a crank? No. I consider a priest somebody who genuinely believes the ideas they're selling, not somebody intentionally deceiving people in order to collect donation money. For this reason, using the words "cranks" and "idiots" is probably not likely to work - something like "If you don't bother to support your points with rational arguments and don't update and keep bothering us, we'll boot you." would be more likely to help them realize it's targeted at them.