Rng renames: gen_ → random_

[dhardy]

• Added a CHANGELOG.md entry

Summary

This appears to be the conclusion of #1503:

rng.random()
rng.random_iter()
rng.random_range(..len)
rng.random_bool(0.2)
rng.random_ratio(2, 3)
rng.sample(Open01)
rng.sample_iter(Open01)
rng.fill(&mut buf)

Opinion

My personal feeling is that this is a poor choice. We recently renamed rand::distributions to rand::distr because the latter was easier to type and consistent with rand_distr (see #1381). This achieves some level of consistency (excepting that random_bool describes its output while other random* methods describe their input), but is not concise.

[newpavlov]

Maybe it's better to do a deprecating backport and release in the v0.8 branch and remove these methods completely in v0.9?

[newpavlov]

I think this name could be misleading. It sounds as if it generates a "random ratio". But I don't have a better proposal than much longer random_bool_with_ratio or the generic argument proposal, so I guess it's fine.

[vks]

I don't really like the new name. I would prefer something like sample_* or uniform_*, but I guess this is too technical.

I agree that random_ratio is misleading, but so is gen_ratio. I think bool_from_ratio or bool_from_prob would also work.