Having a lexicographically sorted collection is a big benefit (I asked GvR years ago to add one to Python but it was no then and seems to be no now!).
Anyone looking for a such a collection could easily miss the critbits model because very few people have heard of them (according to Wikipedia most people in a ration of approx 750:1 know them as radix trees: https://en.wikipedia.org/wiki/Talk%3ACrit_bit_tree).
This was done because partition is an uncommonly used proc that is still
useful in rare cases. There is also a desire to add more procs to this
module at a later time.
