Okay, after further research, I concluded that it's not weird, after all. Here's the explanation: Some search engine crawlers/spiders identify themselves as 'regular' browsers (Firefox, IE7, etc) on existing operating systems (Linux, Windows, etc), thus tricking the tracking engine into thinking that they are not spiders (browser type is set to zero).
On the other side, though, they don't carry cookies, so WP SlimStat can't assign them a visit ID, and there's no 'visit' associated to these pageviews.
So in the end there will always be more 'unique IPs' than visits, because the former also counts the search engine IP addresses, while the latter only identify 'real' human visitors.
But I've come up with a workaround which will address the issue. I've already added it to 2.8.3.
Contact me if you would like to test the new engine.