Improve Optics.Fold docs

[adamgundry]

At the moment Optics.Fold uses foldVL as the canonical introduction form, and has lots of canonical elimination forms. I think it would be clearer to use folding @[] instead for the introduction form, and toListOf for the elimination form. This makes explicit the connection between Fold s a and s -> [a].

[arybczak]

IMO foldVL is much easier to grok than folding 🤔 Its resemblence to traversalVL is analogous to how traverse_ and traverse differ.

[adamgundry]

You're right that we should emphasise the relationship between traversalVL and foldVL. That's helpful if one understands traversals already. But I think that as a route to understanding folds without assuming an understanding of traversals, s -> [a] is simpler.

I'm currently working on a workshop about optics for ZuriHac, which is why I'm thinking about this. Once that is under control I'll try to make a PR with the doc changes I envision.

[phadej]

foldring with a stricter signature, i.e.

(forall b. (a -> b -> b) -> b -> s -> b) -> Fold s a

is IMHO simplest introduction form. Often quite optimal as well.

As I hint in #427, our type for foldVL is not entirely correct - and we don't discuss that in the docs:

*Optics.Core> :t foldVL traverse
foldVL traverse :: Traversable t => Fold (t a) a

works, but i'm not sure it should or shouldn't. (folding foldMap of foldring foldr would be better, wouldn't it?)

[arybczak]

folding foldMap of foldring foldr would be better, wouldn't it?

These provide different performance characteristics. Last time I checked foldVL traverse_ was the fastest for most standard containers.