InnoDB Insert Buffers require indexes that are not unique, why?

The database does not determine the uniqueness of the inserted record when writing to the Insert Buffer. If you look it up again, it’s going to be discrete reading, so InsertBuffer is meaningless. Why is it discrete reading again? For example, if the secondary index named name is inserted (‘kli’, XXX, XXX), then the secondary index is inserted (‘jzou’, XXX, XXX) and then the secondary index is queried again. And the two results are probably no longer in an index page, but are read discretely. IO twice. It must be very inefficient.