More records are always broken during the early years of a data set. There are less other months to beat.
As the data set gets longer, if the climate was stable, the time between records being broken should get longer and longer as each new warmest or coldest month raises the proverbial bar.
It would be interesting to run the same analysis but with the climate data in reverse, i.e. make 2007 the first year and 1659 the last year and then compare the shape of the two graphs.