Consider incorporating datasets and config results from ZAP Benchmark

[Innixma]

Consider integrating the data and results from ZAP Benchmark: https://github.com/automl/zero-shot-automl-with-pretrained-models

• We should be able to run SingleBestConfigScorer rather easily if we prepare the data properly
• We could theoretically run the datasets on AutoGluon AutoMM and compare performance relative to what is recorded in ZAP.

Read their blog for more details: https://www.automl.org/4428-2/