Was there an explicit in-Universe explanation for why Omnius never bothered to physically destroy humans before Tio Holtzman and Sorceresses of Rossak developed the means to defend/attack?
It seems that prior to the events in The Butlerian Jihad, humans would have been very easy to destroy given the full resources of the machines and cymeks (or even just the machines).
Omnius suffers 2 fatal flaws
He keeps subjugated humans because they are useful in small ways, if only for amusement value. His own sick and twisted enjoyment is not unique amongst the AI's; Erasmus also finds torturing humans to be fun, and the implication is that other AIs do so as well.
He fails to see the threat of their creativity until it's too late due to his own overconfidence blinding him to the threat.
There is also a potential third fatal flaw: failure to report. The AI's are, as portrayed, the mirror mental image of mankind's worst individuals... As we see with Erasmus, they don't report every detail, and some even do partial self-wipes to prevent being reabsorbed by Omnius. In short, they suffer the same kinds of flaws as humans. This includes not reporting failures, misrepresenting failures' causes, blaming others for failures, and other forms of misrepresentation.
This means it is quite possible that the real threat wasn't reported out of fear of Omnius. At least not until it was too late.
IIRC, Omnius did not believe in wasting anything. That is why humans were not exterminated and were used as slaves on the synchronized worlds. So by the same token Omnius must have been looking to conquer the free human worlds, rather than exterminate their population.
