From MLP to NeoMLP: Leveraging Self-Attention for Neural Fields