array: generalize S.takeLast and S.dropLast

[davidchambers]

The implementations are not as efficient as they could be, but I want to avoid spreading complexity currently confined to _takeDrop.

[Bradcomp]

I like this because it's simple and straightforward. It looks good to me.

One thing to note is that by generalizing these functions we are creating two partial functions. There is an implicit type requirement that the Foldable is also a finite structure. If we give it an infinite structure the program won't halt, as you can't reverse an infinite structure.

This concern is mitigated for me due to Z.reverse already existing, and because if you do pass a potentially infinite structure to one of these you're probably already doing something wrong. It is more of a theoretical concern.

[masaeedu]

If I'm right about the applicative constraint not being required in #627, then we probably don't need it here as well.

[davidchambers]

You have raised a good point, @Bradcomp; I had not considered infinite structures. I would not expect these functions to work on infinite structures, so I do not find the behaviour surprising, but it is a shame that the type signatures do not accurately convey the constraints.

#627 (comment) applies here too, @masaeedu, so I will merge this pull request in its current form. If it turns out that I have misunderstood the constraints, we can relax the constraints on S.take, S.drop, S.takeLast, and S.dropLast in another pull request. :)