Currently remembered set sampling for young, humongous and old regions is done in three different places, and notification about new data in another place.

Try to reduce the littering a bit.