Make cross-modal benchmark scoring explicitly memory-level

[brianmeyer]

Summary

• split benchmark evaluation into explicit memory-level and raw asset-level scoring
• keep the existing top-line benchmark metrics memory-centric for compatibility
• add asset-level metrics to stage payloads and per-query results for clearer analysis

Linked work

• REC-172
• advances REC-160

Verification

• pytest -q tests/test_cross_modal_benchmark_defs.py -q
• pytest -q tests/test_search_pipeline.py tests/test_config_tools.py tests/test_json_compliance.py -q
• pytest -x -m 'not live' --tb=short

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.