fix(agents): accept Primary-hot warmup KV path

[FluffyAIcode]

Summary

• accept either a complete allens remote hit or an existing Primary local hit during the warmup pass
• retain strict reused-token, zero-compute, zero-fallback, zero-job-failure requirements
• continue requiring a Primary local hit for actual inference
• print compact warm/actual counters when a future gate truly fails

Root cause

After the first successful promotion, repeated deterministic prompts hit Primary on both warmup and actual passes. The gate introduced in PR #185 handled exact allens hits but still required remote_hits >= 1, incorrectly rejecting the hotter local_hits=1 warmup path.

Observed counters were local_hits +2, tokens_reused +80, tokens_computed=0, remote_jobs=0, and fallbacks=0.

Tests

• Primary-hot, allens-hit, and remote-compute gate paths
• diagnostic counter output
• targeted suite — 14 passed
• full platform-neutral suite — 860 passed, 1 skipped
• Python compile and diff checks pass

Made with Cursor

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.