A criticism of the Gram‐Schmidt Orthogonalization process applied to spherical harmonic analysis