Online Gaussian Process Learning Based Adaptive Safe Actor‐Critic Control for Continuous‐Time Nonlinear Systems | Synapse