fix: preserve full document chunks

[suguanYang]

Summary

• publish the full parsed chunk list to document/retrieval state during job finalization
• remove the redundant duplicate media GC and overlap-stat collection path entirely
• keep temp file ownership with the worker task workspace cleanup
• update the parse contract so result ZIP, job_chunks, and document_chunks preserve all parsed chunks and no overlap metadata is written

Closes #31

Tests

• uv run pytest apps/worker/tests/contract/test_parse_task_contract.py::test_should_export_full_result_when_publication_deduplicates_existing_chunks
• uv run pytest apps/worker/tests/contract/test_parse_task_contract.py
• uv run --all-packages --group lint ruff check apps/worker/app/core/tasks/kb_tasks.py packages/shared-python/shared/services/retrieval/publication_service.py packages/shared-python/shared/services/job_lifecycle_sync.py apps/worker/tests/contract/test_parse_task_contract.py
• uv run --all-packages --group typecheck pyright --project pyproject.toml apps/worker/app packages/shared-python/shared