Create a synthetic table where rows have different order according to different distance functions

[Ngalstyan4]

This will make sure that when we use one distance metric but we should in-fact be using another one (e.g. because the user supplied a custom distance metric), then a regression test failure gets triggered.