Harvest code.json files

What: Add support for harvesting federal agency code.json files (Federal Source Code Policy format) into the Data.gov catalog. Add new schema type "code.json", transformation logic to convert code.json releases to DCAT-US 3.0 datasets, validation for both code.json structure and resulting DCAT output, and UI updates to support creating code.json harvest sources.

Why: Federal agencies are required to maintain code.json files cataloging their open-source projects. Harvesting these enables discovery of ~10,000 federal repositories through Data.gov, compliance monitoring for the Federal Source Code Policy, and unified search across both datasets and code. Example: DHS publishes 50+ repositories at `[https://www.dhs.gov/xlibrary/assets/digital-strategy/code.json`—after](https://www.dhs.gov/xlibrary/assets/digital-strategy/code.json) harvesting, each becomes searchable by name, programming language, license, and agency.

Done When:

- [x] Add "code.json" to SCHEMA_TYPE_VALUES in shared/constants.py + database migration
- [x] Create harvester/utils/codejson_mapper.py with codejson_release_to_dcat() function mapping code.json fields to DCAT-US 3.0 (repositoryURL → identifier, name → title, languages → theme, etc.)
- [x] Update harvest.py: handle code.json in __post_init__(), extract releases in acquire_minimum_external_data(), add Record.transform_codejson() method, use repositoryURL as identifier
- [x] Create harvester/utils/codejson_validator.py to validate code.json v2.0.0 structure (dual validation: code.json + transformed DCAT)
- [x] Add "code.json (Federal Source Code)" to schema type dropdown in harvest source form with help text
- [x] Write tests: unit tests for mapper (basic mapping, missing fields, email formatting), integration tests for full harvest workflow (mock code.json, dataset creation), e2e test with real DHS code.json