fix: Filling missing metadata for leaderboard release

[imenelydiaker]

Related to issue #1886.

[x-tabdeveloping]

Made a couple of comments as licence and date are missing some places.
I'm struggling to figure out why the tests are failing though.

[x-tabdeveloping]

Do we not have any information on the dates?

[imenelydiaker]

No, actually it's hard to infer the dates. I assumed you only needed domains so I filled them in pirority.

[x-tabdeveloping]

Also, remember to run linting :D

[isaac-chung]

Looks like tests are failing due to (old?) metadata - Pydantic validation:

sample_creation
  Input should be 'found', 'created', 'human-translated and localized', 'human-translated', 'machine-translated', 'machine-translated and verified', 'machine-translated and localized' or 'LM-generated and verified' [type=literal_error, input_value='derived', input_type=str]

[imenelydiaker]

@x-tabdeveloping I tried to fill a maximum number of missing metadata for the tasks you listed, I used mostly the data we put in the paper.

I don't have the date value for all of them as it's hard to find/infer. I thought it was not a critical metadata to release the LB? If it's not criticial, we can open a good first issue so that people can help filling what's missing.

[KennethEnevoldsen]

Thanks for this fix @imenelydiaker - added annotations from #1910

[KennethEnevoldsen]

Added a few additional annotations (financial where missing and filled out the ArguAna for its Polish translation as well). With this I believe it is good to merge

[x-tabdeveloping]

Thanks for the work @imenelydiaker :D