Fix examples: remove unreliable Cebuano, use clear-cut Filipino

Update Space for combined model

Deploy XLM-RoBERTa hate speech classifier