maskgen: diffloss input parameters do not have text_condition

[gjhhust]

The code on line 564 in https://github.com/bytedance/1d-tokenizer/blob/main/modeling/maskgen.py

 loss, loss_dict = self.diffloss(target=target, z=z, mask=mask, text_condition=text_condition)

However, the forward input of diffloss only has three parameters: def forward(self, target, z, mask=None):