[ENG-8038] fewer blobs in database

[aaxelb]

mainly:

• delete RawDatum model
• knit the rest back together
  • update trove.digestive_tract
    • /trove/ingest view now calls digestive_tract.ingest (which extracts and saves the given record) instead of digestive_tract.swallow (which would save the request body as RawDatum for later extraction)
    • task__extract_and_derive removed (extraction should now be synchronous)
  • (update: moved to [ENG-8221][ENG-8038] less database data; part 1 #877 ) add expiration_date to base ResourceDescription
    • replaces RawDatum.expiration_date
    • add data migration to copy expiration dates from RawDatum to SupplementaryResourceDescription
      (expiration_date is used by osf only for supplements)
  • update SourceUniqueIdentifier.get_date_first_seen
    • use ArchivedResourceDescription.created instead of RawDatum.date_created
• let non-successful CeleryTaskResults get deleted eventually (but keep them longer than successes)
  • CELERY_RESULT_EXPIRES, used for successes, now set from environment variable (with previous hard-coding as default)
  • FAILED_CELERY_RESULT_EXPIRES new environment variable, for all non-success statuses, default 11 days (different prefix because it's not used by celery directly)

also, along the way:

• small fixes, as encountered
  • error at share.example/api/v2/ root -- fix rss/atom feed link generation
  • pagination at share.example/api/v2/sourceconfigs/
  • instead of hard-coded default admin username/password (and undocumented one-off env var)...
    • move env-var access to settings/environment:
      • SHARE_ADMIN_USERNAME, default "admin"
      • SHARE_ADMIN_PASSWORD, no default except when DEBUG
• delete some dead code
  • share.exceptions
• rename IndexcardRdf (and kids) to ResourceDescription, to better convey intent (got confused multiple times by my own bad naming) by parallelling existing ResourceIdentifier
  • (even if "rdf" weren't foul arcana, the usage is wrong/confusing -- "this is a resource description" more helpful than nonsensical "this is an indexcard rdf")
  • move from trove.models.indexcard to trove.models.resource_description
  • ArchivedIndexcardRdf => ArchivedResourceDescription
  • LatestIndexcardRdf => LatestResourceDescription
  • SupplementaryIndexcardRdf => SupplementaryResourceDescription
  • indexcard_rdf => resource_description
  • indexcardrdf => resourcedescription
  • latest_rdf => latest_resource_description
  • latestrdf => latestresourcedescription
  • update_rdf => update_resource_description
  • update_supplementary_rdf => update_supplementary_description
• move more models to their own files (while touching share.models.ingest and trove.models.indexcard)
  • move DerivedIndexcard to trove.models.derived_indexcard (from trove.models.indexcard)
  • leave Indexcard the only model in trove.models.indexcard, appropriately
  • move Source to share.models.source and SourceConfig to share.models.source_config (from share.models.ingest)
• change github actions to stop on first failure (could consider separate jobs for short-circuit and full-run, in future)

ENG-8038

[coveralls]

coverage: 81.714% (+0.5%) from 81.24%
when pulling 85bf0ad on aaxelb:feature/eng-8038--less-blobs
into f0e657b on CenterForOpenScience:develop.

[felliott]

LGTM!

[aaxelb]

may be splitting this across a few deployments -- will return to this and resolve conflicts after #877