Set.fromAscList and Set.fromDescList keep the first of duplicates, inconsistent with other functions

[meooow25]

λ> import qualified Data.Set as S
λ> import qualified Data.Map as M
λ> import Data.Semigroup (Arg(..))
λ> M.fromList [(Arg 1 2, 2), (Arg 1 3, 3)]
fromList [(Arg 1 3,3)]
λ> M.fromAscList [(Arg 1 2, 2), (Arg 1 3, 3)]
fromList [(Arg 1 3,3)]
λ> M.fromDescList [(Arg 1 2, 2), (Arg 1 3, 3)]
fromList [(Arg 1 3,3)]
λ> S.fromList [Arg 1 2, Arg 1 3]
fromList [Arg 1 3]
λ> S.fromAscList [Arg 1 2, Arg 1 3]
fromList [Arg 1 2] -- ??
λ> S.fromDescList [Arg 1 2, Arg 1 3]
fromList [Arg 1 2] -- ??

We should change this to be consistent.

[meooow25]

Unlike Map, Set documentation for any of the fromList functions does not specify which element will be kept in case of duplicates. If we make it consistent, we should also mention it in the docs.

Edit: In fact Map doesn't specify which key will be kept, only the value. But I suppose it would be weird if those did not match.

Are there any advantages in keeping the behavior unspecified? ...I can't think of any.