Missing `item_id` field in LSF Dataset examples

[sergey-serebryakov]

Hi!
I just tried to run the LSF evaluation script (using instructions in README file). I installed project from sources (hash: b4d4c9f, python: 3.11.14). The script failed, and the error was that the time series example (dictionary) was missing the item_id key. I ran it for the weather dataset, so I slightly updated the __iter__ method of the LSFDataset class:

# ...
elif self.mode == "M":
    yield {
        "target": self.data.transpose(1, 0),
        "start": self.start,
        "item_id": self.dataset_name  # This is new
    }
# ...

and I think this fixed it (still running). Is this the right way to fix it, or should I maybe install older versions of some (transient) dependencies?
I'll be glad to create a pull request for it.
Thanks!