if input of these algorithms models AssociativeSequence, then output will model AssociativeSequence, because:
erase and erase_key use: iterator_range, joint_view
pop_back and pop_front use: iterator_range
iterator_range and joint_view preserve associability
if input of these algorithms models AssociativeSequence, then output will not model AssociativeSequence, because:
push_back and push_front use: single_view, joint_view
insert and insert_range use: iterator_range, single_view, joint_view
single_view does not model AssociativeSequence
can fusion let single_view model AssociativeSequence?