When you start to measure the outcomes of the conflict in number of bodies, you cross the line.
And hey... redemption?
The more important point is recognizing the myth that a open conflict with an escalated loss of life will actually end the bleeding by resolving the conflict and the loss of life. Arguing that 'a short open conflict will cost fewer lives than a long unpleasant peace' rests on an assumption that (a) it will be a short conflict, and (
that the rate of loss of life after the conflict will be significantly lower than the loss of life before it.
Looking into history, that's, ahem, remarkably unlikely. The only comparable measure of a thousand years is another thousand years: not just the one, ten, or hundred years of fighting and skirmishing 'openly', but also the rest of the millennia after that.
It's completely impossible to prove the results of course, but given the implied scale of the Circles (a few hundred at most), the unique rate of death for the Circles is already pretty low, and easily below a single modest war. Round up the 17 Annullments to 20, and place them at 500 per Circle just to be on the high side, and over the better part of a millenia you'd have 10,000 people... which sounds impressive, until you remember time scale.
If you spread that over a millenia (remember, nice rounding), that would be 10 per year. 10,000 total.
Compare that to, say, the Iraq War, where the low estimates put the death toll at averaging over 11,000 a year over ten years (2003-2013). In the American Revolution, 25,000 Americans are believed to have died during about 8 years, not including almost 20,000 British sailors and even more British soldiers. The War of 1812, an even lighter affair, has estimates of 15,000 over three years dying from all causes (less than 4,000 from combat on both sides). If you want to go back to more primitive, less capable conflicts, the First Jewish-Roman War (the one commonly associated with 'make a desert and call it peace'), was between 400,000 and over a million over just 7 years.
Really, just peruse this wiki page.
Point is, 10,000 over a millenia is small change. 10,000 deaths (deaths, I repeat: not casualties including wounded and sick) is the scale of modest wars of relatively comparable powers over periods of time less significantly less than a decade.
It shouldn't need to be noted how many of these modest, small conflicts had successor conflicts of similar or even greater scale within decades.
So the next time someone argues to you that a short war will resolve a problem and cost less than a long peace... be skeptical. The political conflict-resolving properties of war are notoriously overrated, and not only do they tend to have another conflict in short order as often as not, but even small and short wars can easily accumulate far more deaths than long-help policies of peace.