Half-precision floating point (16-bit) implementation of the trigonometric function Sin for inputs scaled by pi