feature/mx-1352 Revamp primary source loading

[cutoffthetop]

• fixing typo in extract_oranigram_units (now: extract_organigram_units)
• remove dependency to sqlalchemy
• remove BaseSettings.sqlite_path
• hardcode the stableTargetId of the mex primary source to resolve cyclic dependency
• remove sqlite-based logic
  • insert_primary_source_into_db function
  • seed-mex-db script
  • insert_test_primary_sources_into_db helper function
  • seed_primary_sources_into_db fixture
  • because the middle-step of loading the primary sources into sqlite is now cut out
  • they are read directly from the JSON file in assets where they are maintained
• add BaseSettings.primary_sources_path to point to the primary-source.json file
• replace transform_mex_db_primary_source_to_extracted_primary_source with transform_seed_primary_sources_to_extracted_primary_sources which operates on iterables
• add extracted_primary_sources fixture that loads all primary sources from primary_sources_path and returns a mapping of human-readable identifiers to ExtractedPrimarySource objects
  • this removes the need for creating individual extracted_foo_primary_source fixtures

[mr-kamran-ali]

Looks good other than poetry conflicts.