The objective of this ticket is propagate reference-risk and reference-need model scores as part of our base article.

To do

• Update scores schema. (Refer to RFC). Note that for reference-risk, we want the References in go struct, but suppressed in json.
• Update documentation/json-schema
• Update wmf submodule to accommodate reference-risk and reference-need API call. (Refer to RfC). Note that for reference-risk, we want to call with extended_output parameter.
• Update structured-data: articleupdate to call revertrisk, ref risk and ref need in parallel
• Call all score APIs with a configurable deadline provided by env (Start with 1 second)
• Handle API errors and timout by simply logging. Publish the event without score.
• Update schema on structured-data, ondemand, realtime, main API, structured-contents, snapshots
• Create new migration on realtime for ksqldb stream
• Deploy to dev
• Monitor cloudwatch logs after a week. Make note of unexpected API errors. (We will follow up with ML for these)

QA (acceptance criteria)

• Able to see all 3 scores in aws.strucutred-data.articles.v1 topic
• Make on-demand API calls from dev for the articles above - we should see all 3 scores, however, references must be omitted from reference-risk output.
• Check newly generated snapshots on dev. Articles should have all 3 scores, however, references must be omitted from reference-risk output.
• Connect to dev realtime. Articles should have all 3 scores, however, references must be omitted from reference-risk output.