test: test by comparison to naive impls

[laggycomputer]

• test: naive theta pi
• feat: naive watterson theta

[molpopgen]

Are we unable to determine the ploidy from the VCF file itself? What happens if we get the value wrong during input?

[laggycomputer]

Looks like the VCF standard never defines a place where ploidy is specified and mixed ploidy is supported as well. Not sure we can do anything.

[molpopgen]

I would argue to drop these traits. I think we want the naive versions of each stat to be defined somewhere (preferably where they cannot access fields of the GenomeCollection), and have a definition like:

pub fn diversity_from_sites(iter: Iterator<Item=&Site>) -> f64 {todo!()}
pub fn diversity_from_samples(iter: Iterator<Item=Sample>) -> f64 {todo!()}

[laggycomputer]

GenomeCollecction.data is private so implementors already can't access it. We can redefine it in terms of iterators though, yes.

[molpopgen]

GenomeCollecction.data is private so implementors already can't access it. We can redefine it in terms of iterators though, yes.

But you have the naive GlobalPi in the same source file, which brings up the potential to access internal fields.

[molpopgen]

One other comment: the conventional commit syntax for this pr should be test: rather than feat: when it comes time to merge. Only tests are affected and not the public API.