Probably also because most token use cases are not latency sensitive. A 200ms extra delay isn't going to change much for most use cases.