fix(autoresearch): deploy current worker before proposal

[FluffyAIcode]

Summary

• deploy and verify the current candidate before invoking the real Gemma Strategy Agent
• clear Primary cache after predeployment
• keep proposal generation inside the hard-fail restore boundary
• redeploy and clear again after a proposed candidate changes runtime configuration
• add visible lifecycle phase output to the original supervisor Terminal

Root cause

The first real supervisor called Strategy Agent Prefill before any deployment. allens was still running a stale pre-window-aware process and rejected the prompt with the recurring 1.0924GB estimate. Proposal happened outside the try/restore block, so no preflight remediation ran.

Tests

• source-order contract verifies deployment precedes proposal
• lifecycle phase output contract
• supervisor candidate/results/baseline tests — 6 passed
• full platform-neutral suite — 889 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.