Raij\=u: Reinforcement Learning-Guided Post-Exploitation for Automating Security Assessment of Network Systems