tl;dr
The title says it all.
Prologue
One of the keys to my personal learning is that, very early in my database career, I taught myself how to make lot’s of rows of Random Constrained Data in a comparatively short time. With the help of a few good folks over time, the method has been tweaked and improved for performance.
It used to be sufficient to generate a million rows of test data to prove some point, especially where performance in code is concerned. With today’s machines, though, it’s gotten to the point were 100 million rows almost seems inadequate.