IRV (better to use the general class of methods called “Condorcet”) has distinct advantages, it is true. That process also has limitations. See the rather well-written analysis at:  wiki.electorama.com/wiki/Condorcet_method ;
  http://wiki.electorama.com/wiki/Condorcet_method .

For 11 years, I have participated in a local group that works on the integrity of election systems, the WFV. It started, in about 2000, with just the analysis that you made, John. (Plus a consideration of fair representation in Districts.)

In short, IRV can indeed do what you note - bring a consensus candidate to the fore, within a highly contested multi-candidate race.

But there are snags and difficulties, and both you and I support a fair and balanced consideration of both sides of a public proposal, before declaring a conclusion.

One of the downsides is voter-education, so that they really mark their multi-choice ballots properly - both in the sense of who they want to elect, and in the mechanics of the markings so the system properly reads and registers the true voter intent.

Another negative is higher costs, as generated by the complexity, in both equipment and staff time.

One negative that would really bother people (because it is so very visible) is the double-barrelled case of the delay for final results, and the public outcry if the candidate who got reported as the early leader, is suddenly dumped and another one, who was running behind, is declared the winner! Put in your earplugs! That delay happens since the vote tally process can only be properly done with ALL the ballots in hand and “on the table”, so that it is known whether the second-preference process will need to be done. Ballots from many voters, especially those from overseas military, etc, may delay being processed for many days.

Then there is the process of bringing all the ballots into one single master dataset, to run the comparison process. Within one County - not too hard - we already do that for Whatcom, and probably all WA counties. All ballots are run thru the same counting process at the same place, a necessity for Condorcet.

But for Legislative Districts, the system would require determination of some central processing facility, which would consolidate all the ballots from multiple counties in order to do the rankings. Who would run that? And bigger yet - in what way would the “ballots” be sent from each outlier? Both sending methods have serious snags—if the physical paper ballots are transported, serious questions of security and confidence arise, plus the lesser question of where to put them once it is over? If the data is transmitted electronically, serious questions of security and confidence arise! (Ooops, we just said that!)

And, don’t even imagine thinking about doing State-wide races.

So, the real-world will not make Condorcet systems likely methods, in spite of their plusses.

Oh, maybe our winner-rules should be changed from Top-Two, to a Top-Three?