Columnable for newtype

[guibou]

Is your feature request related to a problem? Please describe.

I'm working with a lot of newtype on top of numeric types, (e.g. Double, Int, ...), and I recently got confused by fromAny which does not behave as expected with them, e.g, it returns Nothing.

ghci> Right df <- fromCsv "a,b,c\n1.3,2.5,7.2"
ghci> newtype BetterDouble = BetterDouble Double deriving (Show, Eq, Ord)
ghci> fromAny (snd $ head (head (toRowList df)))  :: Maybe Double
Just 1.3
ghci> fromAny (snd $ head (head (toRowList df)))  :: Maybe BetterDouble
Nothing

Describe the solution you'd like

Having a way to either extend, or add support for anything coercible to Double

Describe alternatives you've considered

For now, I'm using fromAny @Double followed by coerce, but I got surprised and I don't like surprise ;)

Additional context
Add any other context or screenshots about the feature request here.

[mchav]

fromAny is sort of like Data.Dynamic in that it checks that the type you're trying to get out is the type stored inside. In fact, the only reason I didn't use Data.Dynamic is because I needed to import the Columnable constraint.

So for example:

ghci> newtype BetterDouble = BetterDouble Double deriving (Show, Eq, Ord)
ghci> import Data.Dynamic 
ghci> x = toDyn (1 :: Double)
ghci> :t x
x :: Dynamic
ghci> fromDynamic x :: Maybe Double
Just 1.0
ghci> fromDynamic x :: Maybe BetterDouble 
Nothing

Also FYI custom number types make things a little slower since it would be stored in a boxed vector (our unboxable universe is closed) so native doubles will be a lot faster - about 3-4 times faster.