[Bug] Evaluation result is being overwritten

[marmg]

Summary

In the evaluation, we update a result dict with the metrics. However, if both linker and mentions_extractor are defined the result will be overwritten:

if nlp.get_pipe("zshot").linker:
    pipe = LinkerPipeline(nlp, batch_size)
    result.update(
        {
            field_name: {
                'linker': linker_evaluator.compute(pipe, dataset[split].select([1, 4]), metric=metric)
            }
        }
    )

if nlp.get_pipe("zshot").mentions_extractor:
    pipe = MentionsExtractorPipeline(nlp, batch_size)
    result.update(
        {
            field_name: {
                'mentions_extractor': mentions_extractor_evaluator.compute(pipe, dataset[split].select([1, 4]),
                                                                           metric=metric)
            }
        }
    )