docs: clarify context-efficiency evidence

[Coding-Dev-Tools]

Summary

• make the README's reproducible context-efficiency evidence easier to find and interpret
• align the documented performance command with the 10-iteration, 260-recall measurement it cites
• include the approved readable benefit-flow image and compact the README while preserving tested privacy, licensing, and Cloud Sync disclosures

Evidence

• long-document fixture: 73.0% less retrieved context at Recall@5 1.000
• compact MCP fixture: 55.38% smaller serialized response with unchanged retrieval scores
• each measurement has a separate boundary and is not a billing, latency, customer-savings, or external-benchmark claim

Validation

• python -m pytest tests/ -q
• ruff check .
• python -m eval.harness --dataset eval/datasets/codemem.jsonl --k 5
• python -m eval.chunking_eval
• python -m eval.performance --dataset eval/datasets/codemem.jsonl --k 5 --iterations 10 --json
• python scripts/check_commercial_manifest.py

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9bad370813

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".