Steve, the IRV method of counting rank choice ballots almost never elects a majority winner out of all persons who cast votes in an RCV/IRV election. It simply elects the candidate that a majority of voters whose ballots have not been exhausted (whose ballots are still being counted) in the final counting round. An analysis of IRV/RCV elections in California found the normal case is for the IRV method of counting RC ballots to elect a winner with significantly fewer than 50% of all votes from the original voters - usually somewhere in the 30 to 45 % of voters decide the final counting round because IRV method often eliminates many voters' subsequent choices prior to when their prior choices are counted.
I.e. in RCV/IRV a voter's lower ranked choices can never hurt their higher rank choices but a voter's higher rank choices commonly hurt the chances of winning of their lower ranked choices. It's part of what causes non-monotonicity.
That said, I do not know the specific method NYC is using to pass any remaining lower choice votes when the upper choices are eliminated, so I've forwarded your question to another person who is more involved in what's happening in NYC.