pub const fn sin(x: Real) -> Real
sine approximation.
Maximum absolute error ≈ 4.44 × 10^{-16} (≈ 2 ULP near |sin| = 1).