Add support for models using final_logits_softcapping to AsyncGRPOTrainer

### Feature request Add support for models using final_logits_softcapping to AsyncGRPOTrainer ### Motivation Some models (like gemma 2) use final_logits_softcapping and currently cannot be run with AsyncGRPOTrainer. ### Your contribution I would be happy to collaborate on this feature request by creating a potential PR.