I think the correct way to handle this is to include a bad-bot blocker in your webserver. There are plenty scripts and addons available for the common software stacks.

There are also local and privacy-respecting Web Application Firewalls like ModSecurity, Janusec, Vulture Project (I haven’t yet tested them) which could maybe do the same thing.

We’re all subject to these crawlers, bots and vulnerability scanners. I also run 3 small websites including mail and a few other services. I rarely block some bot that downloads images over and over again. And fail2ban blocks a lot of brute-forcing attempts. Other than that, the traffic they cause isn’t that much compared to a single other service like Matrix chat or some Fediverse software that causes lots of HTTP requests all day long. It runs without Cloudflare or other third-party services for years on my slow home internet connection. Back then even on a single board computer (like the Raspberry Pi.)

So my experience is a bit different. And that I can run 3 websites on a RasPi on a 15MBit connection just fine and other people need Cloudflare for a 1000MBit VPS makes me think it’s snake-oil. But yeah, I agree if you block the bots, they stop after some time. That’s also my experience. But the traffic isn’t that much in the first place and there are better ways to do it in my opinion.