[ENH] compare models with and without weight

[AdolHong]

sometimes we want to compare model with weight to without weight, but, because the val metric is multiple weight, this multiplication make models cannot be compared.

you can change code below:

file: pytorch_forecasting/models/base_model.py , line 665:

def validation_step(self, batch, batch_idx):
    x, y = batch
    if y[1] is not None:
        weight = torch.where(y[1]!=1.0, torch.tensor(1.0), y[1])
        y = (y[0], weight)
    log, out = self.step(x, y, batch_idx)
    log.update(self.create_log(x, y, out, batch_idx))
    self.validation_step_outputs.append(log)
    return log

[fkiraly]

Interesting suggestion - a pull request would be appreciated so we can see the change in the context of the current code.