vLLM implements VaultGemma via Transformers backend

VaultGemma is now supported in vLLM through the Transformers modeling backend. This integration allows for optimized inference of the VaultGemma model architecture.