Space-Time Approximation with Shallow Neural Networks in Fourier Lebesgue spaces