Introduce MSFAsync

[turion]

Work in progress to clean up the module Data.MonadicStreamFunction.Async.

@ivanperez-keera would you in principle merge this? Then I would clean up, and document.

[turion]

I got the basic insight from @AshleyYakeley in haskell/mtl#85 (comment). Since MSF (MaybeT m) a b == ListT (ReaderT a m) b, it should have a monad instance. This PR implements it, and gets the function concatS for free.

[ivanperez-keera]

A couple of comments:

• Can you first create an issue to document what's happening?
• Don't mix fixes to other things as part of the same PR (even if they are in the same module).
• The new type should be by default hidden in the module. It's an implementation artifact.
• How does the new implementation of concatS work? It looks pretty complicated to me. Can you document it and potentially introduce intermediate names (and types) for the subexpressions?

[ivanperez-keera]

BTW we also have a discussion tab were we can start the conversation. One things are clearer we can turn it into an issue. That can help separate long conversations (discussions) from more formal (and succinct) change proposals (issues).

[turion]

Can you first create an issue to document what's happening?

This is #338