Is interpolation benign for regression random forests?