Add v2 page-memory ingestion API

[suguanYang]

Summary

• Adds the /v2 API registry for jobs, documents, and retrieval.
• Routes /v1/jobs through legacy chunk ingestion and /v2/jobs through server-selected ingestion commands.
• Uses page-memory for v2 PDF/PPTX jobs, while accepting other supported v2 file types on the legacy chunk path for v1-compatible behavior.
• Persists canonical job metadata for api_version, resolved parse_track, processing_generation, and page-memory worker config.
• Moves page-memory runtime config out of public request/env switches and into server defaults captured in job metadata.

Closes #194

Tests

• uv run --all-packages pytest apps/api/tests/contract/test_page_memory_parse_track_contract.py apps/api/tests/contract/test_job_creation_contract.py
• uv run --all-packages pytest apps/worker/tests/contract/test_page_memory_asset_java_contract.py apps/worker/tests/contract/test_page_memory_fine_hierarchy_contract.py apps/worker/tests/contract/test_page_memory_node_assembler_contract.py apps/worker/tests/contract/test_document_agent_budget_contract.py
• make lint
• make typecheck
• git diff --check

Local smoke test

• Started local API and worker.
• Created /v2/jobs PDF job job_839fb1c5b274; worker ran parse_track=page_memory; job reached done; /v2/retrieval/query returned a page result.
• Created /v2/jobs TXT job job_a90b35ae17e3; worker ran parse_track=chunk; job reached done; /v2/retrieval/query returned a text result.
• Verified /v2/jobs rejects public parse_track and /v1/jobs rejects parse_track=page_memory.