Transition away from ocd-ids instruction source

[chrismytton]

Problem

Fuzzy matching OCD areas in the build process is causing problems with Uganda, see everypolitician/everypolitician-data#31158 (comment) for more details. The code for handling OCD ids has never been terribly satisfying and seems to be a constant source of issues.

Proposed Solution

Remove the ocd-ids instruction type. Currently it's only Australia/Representatives and Uganda/Parliament that are using this, so perhaps it would be better to make the OCD ids a manually generated CSV for those countries.

We want to try and transition to using MapIt and Wikidata as our primary sources of area information, rather than using OCD IDs directly. Migrating these two legislatures over to one or both of Mapit/Wikidata seems like a better use of our time than fixing the OCD fuzzy matching problems.

Acceptance Criteria

Australia/Representatives and Uganda/Parliament should no longer have an ocd-ids instruction type and should be getting their area information from Mapit and/or Wikidata.

Not Required

We don't need to change anything in other legislatures that include OCD IDs from the scraper, these are fine as they are since the ID is just used as a plain text identifier string. This change is just about changing the two places that try to do matching between area names and OCD IDs.

Related Issues

• Add Mapit as a source Add MapIt as a source #595
• Add Wikidata for Areas Add Wikidata for Areas #567

[tmtmtmtm]

The areas for Australia/Representatives are now mapped to Wikidata (since everypolitician/everypolitician-data#51981). They're still keyed (and reconciled) on OCD ids, but it will now be easier to switch entirely to Wikidata IDs.