...and possibly also for atan2 and hypot, if it is appropriate to move the special casing in ForwardDiff.jl for those functions to this package.
I will submit a PR shortly.
Of course, this would depend on resolving jrevels/RealInterface.jl#4 upstream.