Add Gold Commodity News scenario

[yifanmai]

Gold commodity news is a classification benchmark based on a dataset of human-annotated gold commodity news headlines (Sinha & Khandait, 2019).

[yifanmai]

Hi @ryokawajp, please review this pull request.

By approving, you acknowledge that:

• You will be listed as a co-author of the merge commit
• You agree to license your work under Apache 2.0

[ryokawajp]

Just for confirmation, this scenario "gold_commodity_news" corresponds to "news_headline" in the original source code, right?

[yifanmai]

Yes, this corresponds to news_headline in #3064. This dataset doesn't seem to have a canonical name in the literature, so I would prefer to pick a name that is less ambiguous.

[mtake]

looks good to me

[ryokawajp]

I tested it locally.
news_headline is renamed to gold_commodity_news, and the values of category option is changed to a style like price_or_not from Price or Not. I think these changes are reasonable.

I am not sure about test_ewok_scenario.py.

LGTM.

[yifanmai]

Resolved the merge conflict - it was required to get an existing failing test to pass, but this has since been resolved by #3069.