That's an interesting question, especially since OKRs are gaining in popularity, especially in Silicon Valley. In my opinion there are a couple conflicts between Holacracy and the OKR philopshy. The first is in how OKRs are defined. OKRs assume the existence of a management hierarchy and OKRs are typically defined at the top and trickle down. I suppose you could define OKRs at the circle/role level instead, but as you mention that could become burdensome with most people filling many roles in Holacracy.
The bigger conflict is that the key-results of OKRs are targets to achieve by a certain date. They are an attempt to predict the future and there is an implied commitment that you will achieve 60-70% of the key results. This of course can create extra stress if the commitments are no longer realistic and sub-optimal prioritization if these goals are no longer the most important thing to be doing. Brian explores this topic in much more detail in his article The Insanity of the What-by-When. Definitely recommend you check that out if you haven't seen it.