Mismatch between TokenAttributions model and UI

[dcfidalgo]

Describe the bug
The TokenAttributions model accepts multiple label-attribution pairs, while the UI can only handle one.

To Reproduce
Steps to reproduce the behavior:

import rubrix as rb

record = rb.TextClassificationRecord(
    inputs="mock",
    prediction=[("POSITIVE", 0.9), ("NEGATIVE", 0.1)],
    explanation={"text": [
        rb.TokenAttributions(
            token="mock", attributions={"POSITIVE": 0.5, "NEGATIVE": -0.5}
        )
    ]},
)

Expected behavior
Maybe for now the model should only accept one label-attribution pair, or the UI should be able to handle various.

Environment (please complete the following information):

• OS [e.g. iOS]: Ubuntu
• Browser [e.g. chrome, safari]: firefox
• Rubrix Version [e.g. 0.10.0]: master
• ElasticSearch Version [e.g. 7.10.2]: 7.10.2
• Docker Image (optional) [e.g. rubrix:v0.10.0]:

[dcfidalgo]

@frascuchon @leiyre