Pauli Network Circuit Synthesis with Reinforcement Learning