Add root service query mode

[jioffe502]

Summary

• add retriever query --run-mode service through the query-owned typed request path
• add a small service query workflow and RetrieverServiceClient.query() for POST /v1/query
• preserve root query JSON output shape and apply candidate/page/content shaping client-side after service results
• reject local-only LanceDB/embed/rerank flags in service mode instead of silently dropping them

Stack note

This draft is opened against upstream/main for visibility. It currently includes the current #2221 service-ingest commit plus the service-query commit because the query branch is stacked on that work.

Validation

• pre-commit run --files nemo_retriever/src/nemo_retriever/adapters/cli/main.py nemo_retriever/src/nemo_retriever/query/options.py nemo_retriever/src/nemo_retriever/query/service.py nemo_retriever/src/nemo_retriever/query/workflow.py nemo_retriever/src/nemo_retriever/service/client.py nemo_retriever/tests/test_query_workflow_options.py nemo_retriever/tests/test_root_query_cli.py nemo_retriever/tests/test_service_query_client.py
• PYTHONPATH=/localhome/local-jioffe/NeMo-Retriever.worktrees/service-query-integration-base/nemo_retriever/src /localhome/local-jioffe/retriever-skills/nemo_retriever/.venv/bin/python -m pytest nemo_retriever/tests/test_query_workflow_options.py nemo_retriever/tests/test_root_query_cli.py nemo_retriever/tests/test_service_query_client.py
• git diff --check codex/service-query-integration-base..codex/root-query-service-mode

Live jp20 service validation was previously attempted, but no service was reachable on localhost:7670 or localhost:7671 from this environment.