feat: expose label to create scores API

[desaxce]

Changes:

• expose label to create_scores API

To test:

• create an experiment from Python:

experiment = dataset.create_experiment(
    name="Next 1",
    params=[{ "type": "Mon expérience", "top_k": 2 }]
)

for index in range(2):
    scores = [{ 
        "name": "Sentiment",
        "type": "AI",
        "label": f"happy-{index}", # HERE IS THE LINE TO ADD COMPARED TO CLASSIC EXPERIMENT
        "value": 55
    }]

    experiment_item = {
        "datasetItemId": # Put the ID of a dataset item
        "scores": scores,
        "input": { "question": "kesako?" },
        "output": { "contexts": ["rien"] }
    }
    
    experiment.log(experiment_item)

If you have a platform version greater or equal to 0.0.627-beta, you should see the label instead of the score value in the experiment details:

[clementsirieix]

FYI label is not defined in ScoreDict but it should.
Also it does not seem to be part of any signatures so it's hard to tell if this is intended.

[desaxce]

FYI label is not defined in ScoreDict but it should. Also it does not seem to be part of any signatures so it's hard to tell if this is intended.

Added. label is not very discoverable indeed. DatasetExperiment.log will now have it, but it's hidden behind DatasetExperimentItemDict.scores. We need to find a way that logs an experiment item in an easier way: having to pass scores in the experiment item dict is cumbersome.

[clementsirieix]

lgtm