The RealFloat data type context was dropped from Data.Complex some time
ago (see ea280135). However, the rest of
the API in that module was left as-is even though many of the accessors
in that module would work with much more general types now.

This change simplifies the signatures of the functions above, which in
the current API all unnecessarily take a RealFloat constraint that they
don't use (and which may cause to pass around superflous type-class
dictionaries):

    realPart  :: Complex a -> a
    imagPart  :: Complex a -> a
    conjugate :: Num a => Complex a -> Complex a
    mkPolar   :: Floating a => a -> a -> Complex a
    cis       :: Floating a => a -> Complex a

This was originally proposed by Edward Kmett in

  http://www.haskell.org/pipermail/libraries/2014-March/022358.html



Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>