In Blitz there seems to be a psychological bump at the hundredths starting from 2000, maybe 1900 too.
i'd go look at the stats..
lichess.org/stat/rating/distribution/blitzThe theory is that there is a celebration lull... But it maybe be something more mystical.
Edit: thinking back at the "theory" of the resting after hundreth reach... Since glicko is more data responsive to the actual distribution of game being paired in a dynamical sense (volatility modelling), such a lull would have likely a signal left in the statistics for fast enough time controls. did lichess plot the corresponding deviation statistics, ever?