Add Agent Knowledge OS closeout benchmark

[yvette-carlisle]

Summary

• Add the XY-1023 Agent Knowledge OS closeout benchmark report and JSON snapshot covering 19 products across six scenarios.
• Preserve competitor-specific strengths, including VectifyAI PageIndex/OpenKB, as optimization inputs instead of broad pass claims.
• Wire the report into README and docs/evidence/benchmarking index, with regression tests for key boundaries and docs wiring.

Validation

• cargo make real-world-memory
• python3 -m json.tool apps/elf-eval/fixtures/report_snapshots/2026-06-20-agent-knowledge-os-closeout-benchmark-report.json
• cargo test -p elf-eval --test real_world_job_benchmark agent_knowledge_os_closeout_benchmark -- --nocapture
• cargo test -p elf-eval --test real_world_job_benchmark -- --nocapture
• cargo make test
• cargo make fmt
• cargo make check-docs
• decodex docs check
• git diff --check
• cargo make check