feat(rules): map and validate job export rows

[jaeyeopme]

Motivation

Add the pure data rules needed before output writing and export orchestration.

Scope

• Add source-to-export candidate mapping.
• Add required field, status, and date validation for mapped job rows.
• Add deterministic duplicate handling where the first valid job id wins.
• Add focused tests for mapping, validation, duplicate policy, and MVP rule counts.

Out of Scope

• Reading fixture pages.
• Writing output files.
• Storing run history.
• API or CLI export behavior.
• Public docs status changes.

Modifications

• Add mapping functions that preserve source context and compute latest update timestamps from job and related update records.
• Add validation functions that return accepted clean jobs or rejected rows with documented reason codes.
• Add duplicate handling that rejects later valid rows with the duplicate_record reason code.

Test Plan

• uv run pytest tests/test_mapping.py tests/test_validation.py tests/test_duplicate_handling.py -v
• uv run ruff check src/clean_data_export_api/mapping.py src/clean_data_export_api/validation.py tests
• uv run basedpyright
• uv run pytest tests/test_mapping.py tests/test_validation.py tests/test_duplicate_handling.py --cov=clean_data_export_api.mapping --cov=clean_data_export_api.validation --cov-report=term-missing
• uv run pytest -v
• uv run pytest --cov=clean_data_export_api --cov-report=term-missing --cov-fail-under=90
• uv run python -m compileall src tests

Docs Impact

none

ADR Check

• No architecture or scope decision needs a new or updated ADR.

Result

The core job export rules are deterministic, testable, and independent from file output, persistence, API, and CLI code.