bug in computing scores for longdialogue_qa_eng

[Xianchao-Wu]

https://github.com/OpenBMB/InfiniteBench/blob/main/src/compute_scores.py#L238

1. only one reference label is used for comparison, better loop around each answer in label, e.g., label=['ECKER', 'COMMANDER BILL ECKER'];
2. prediction phrase is splitted into words for comparison, which means: pred=STAMP PAID, label=['STAMP PAID'], score=0.0, data_name=longdialogue_qa_eng

and

pred=ADRIAN, label=['ADRIAN'], score=1.0, data_name=longdialogue_qa_eng

[tuantuanzhang]

We have fixed this bug in the latest commit. Thank you for your notification.