Federated Self-supervised Speech Representations: Are We There Yet?