Deploy linear scan cold-eye prefetch improvements

[leonanos8]

Summary

Updates the production linear-scan shadow deployment to a new image containing:

• Coalesce cold-eye prefetch and bound foreground fetch latency #2367 cold-eye prefetch coalescing, merged through Add cold-eye cache and prefetch for CPU linear scan #2351 → Fused mirror scan, packed pair kernel, and lane pipelining for the CPU linear scan #2348.
• Existing Adjust linear scan image #2358 certificate/image changes.
• SMPC__COLD_EYE_LFU_CACHE_RECORDS increased from 4096 to 12288.

Rollout

Deploy in shadow mode and monitor SuperMatcher spikes, scan latency, cache/prefetch metrics

Out of scope

Network IRQ affinity will be handled separately through node provisioning, in a next step

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-02T11:56:49.469880Z	8f9d085	PR opened

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.