Quantum Kernel Estimation With Neutral Atoms For Supervised Classification: A Gate-Based Approach