Please, move Kleisli HOFs to separate *F type classes, while main class instances should be specialized to what the class instances are for

[Anton-Latukha]

The initial HNix code has a fully open structure.

So all functions become HOFs that thread pass-though some implementation.

During factoring-out the function application from demand - I noticed - that yes - it simplifies the code, but that HOF legacy gets lost.

Having HOFs for everything is nice to preserve.

Had an epiphany. HOF functions move to the top-level. Class instances specialize HOFs to the HNix implementation.
Proper argument ordering (that currently happens) allows:

So force instead of force: forall . HOF it becomes forceF kleisli. The force becomes force = forceF pure. With that, we would leverage all work & code done and would have simultaneously the HOF abilities and the code elegance and I believe the performance. Because for GHC is easy to understand that class MonadThunk: force = forceF pure is constant and so the function is specialized to the class, giving us free to all this.

The #850 works on it also.
In #776 noted that Value methods are HOFs that can form the class instances.

[Anton-Latukha]

With:
#871
#872
#873
#874
#875

this is closed.