Towards Federated Domain Unlearning: Verification Methodologies and Challenges