Add consistency with foldMap law to UnorderedTraverseLaws

[johnynek]

This adds that unorderedTraverse should be consistent with unorderedFoldMap, which seems intuitively obvious, but isn't checked.

Adding this law passes for both the current implementations of UnorderedTraverse (a typeclass I think very rarely used). However, note I just created: #4802 which argues that UnorderedTraverse[Set] should be deprecated because it should be unlawful (although with the current laws it is not).

[LukaJCB]

lgtm!

[satorg]

@johnynek ,
Is there any reason why we cannot say the same about regular traverse/foldMap?
I personally couldn't think of any, but perhaps I'm missing something.
I tried to add this law to TraverseLaws:

def traverseConsistentFoldMap[A, B: Monoid](fa: F[A], fn: A => B): IsEq[B] =
  F.traverse(fa)(a => Const[B, B](fn(a))).getConst <-> F.foldMap(fa)(fn)

and the tests seem passing.

I can file a PR from that excersice if you believe it makes sense.

[johnynek]

Honestly I just assumed that law was already there. I think this should definitely be a law but I wonder if you could prove it from any other laws that exist. I think we should add it.