Sorta representable

As we've discussed, everything `Distributive` should be `Representable`, but maybe the librarian doesn't want to commit to a particular `Rep`. I just realize that there's actually a way to get a sort of middle ground, though it's a bit awkward.

```haskell
class Functor f => Distributive f where
data family RepD f -- Must be a newtype if Representable
tabulateD :: (RepD f -> a) -> f a
indexD :: f a -> RepD f -> a

class (Distributive f, Coercible (Rep f) (RepD f) => Representable f where
type Rep f

tabulate :: forall f a. Representable f => (Rep f -> a) -> f a
tabulate = coerce (tabulateD :: (RepD f -> a) -> f a)

index :: forall f a. Representable f => f a -> Rep f -> a
index = coerce (indexD :: f a -> RepD f -> a)
```

If the module defining `f` chooses not to export the `RepD f` newtype constructor, then the choice of representation is hidden.

The issue contains a proposed Haskell design for Distributive and Representable, including RepD, tabulateD, indexD, tabulate, and index. No files, tests, or entry points are named, so first determine where these classes are defined and whether the proposal has been accepted. Done would require an agreed implementation scope and corresponding validation.

Written by the indexing model from the issue text.