Persist cook phases before Lab dispatch

[chubes4]

Fixes #7891.
Fixes #7892.

Root cause

The controller created its durable Lab proxy only inside exec_lab_context, after workspace materialization and dependency hydration had completed. During those Homeboy-owned setup phases, the caller's cook ID had no lifecycle record and status could not explain the active work or provider policy.

Changes

• Resolve the stable caller-supplied agent-task run ID before workspace staging and pass that same identity through materialization.
• Persist a controller-owned proxy at the start of materialization.
• Update its phase through materializing, hydrating, and dispatching.
• Record runner identity, source checkout metadata, remote workspace when known, configured provider rotation candidates, and truthful provider_state: pending before model execution.
• Preserve the existing fix: rotate stalled agent-task providers #7758 attempt/rotation history and fix: preserve agent task run metadata across Lab offload #7894 runner-child handoff/reconciliation contract under the same caller ID.

How to test

1. Run cargo test controller_proxy_records_pre_execution_phase_progress -- --test-threads=1.
2. Run cargo test core::agent_task_lifecycle::tests -- --test-threads=1.
3. Start homeboy agent-task cook --runner <runner> --lab-only --run-id test-preflight ... against a workspace that requires materialization.
4. While workspace setup is active, run homeboy agent-task status test-preflight --full.
5. Confirm status resolves before provider dispatch and reports metadata.phase as materializing or hydrating, metadata.provider_state as pending, configured metadata.provider_rotation, the selected runner, and source checkout identity.
6. Confirm the same caller-ID record advances to dispatching, binds the runner child after daemon acceptance, and retains provider attempt history if rotation occurs.

Verification

• cargo fmt --check passed.
• cargo check passed.
• New focused test passed: 1 passed, 0 failed.
• Agent-task lifecycle suite passed: 42 passed, 0 failed.
• cargo test --lib: 7,171 passed, 34 failed, 18 ignored. The 34 failures are existing environment/baseline failures across extension discovery, fixture configuration, signal timing, and unrelated deploy/audit tests; the affected lifecycle suite is green.

Backwards compatibility

Additive metadata only. Existing run IDs, runner handoff records, attempt history, and status consumers remain valid. Commands without an agent-task run ID do not create a proxy.

AI assistance

• AI assistance: Yes
• Tool(s): GPT-5.6 Sol via OpenCode
• Used for: Implemented pre-execution lifecycle persistence, provider provenance, and regression coverage under Chris Huber's direction

[homeboy-ci]

Homeboy Results — homeboy

Lint

✅ review lint — passed

ℹ️ Full options: homeboy self docs commands/lint
ℹ️ Save lint baseline: homeboy review lint homeboy --baseline
Deep dive: homeboy review lint homeboy --changed-since 225f2bf

Artifacts and drill-down

• CI results artifact: homeboy-ci-results-homeboy-review-lint-homeboy-Linux contains immediate command JSON for this action invocation.
• Observation artifact: homeboy-observations-homeboy-review-lint-homeboy-Linux contains exported Homeboy run history for deeper queries.
• Drill-down: download the observation artifact, then run homeboy runs import <dir>, homeboy runs list, and homeboy runs findings <run-id>.
• Artifacts are attached to the workflow run: https://github.com/Extra-Chill/homeboy/actions/runs/29223505287

Test

⚠️ review test — baseline red

Deep dive: homeboy review test homeboy --changed-since 225f2bf

Artifacts and drill-down

• CI results artifact: homeboy-ci-results-homeboy-review-test-homeboy-Linux contains immediate command JSON for this action invocation.
• Observation artifact: homeboy-observations-homeboy-review-test-homeboy-Linux contains exported Homeboy run history for deeper queries.
• Drill-down: download the observation artifact, then run homeboy runs import <dir>, homeboy runs list, and homeboy runs findings <run-id>.
• Artifacts are attached to the workflow run: https://github.com/Extra-Chill/homeboy/actions/runs/29223505287

Audit

✅ review audit — passed

• structural — 46 finding(s)
• field_patterns — 17 finding(s)
• core_boundary_leak:core-agnostic-source — 16 finding(s)
• intra-method-duplication — 9 finding(s)
• dead_code — 5 finding(s)
• test_quality — 5 finding(s)
• Types — 3 finding(s)
• near-duplication — 3 finding(s)
• docs — 2 finding(s)
• duplication — 2 finding(s)
• Total: 111 finding(s)

Deep dive: homeboy review audit homeboy --changed-since 225f2bf

Artifacts and drill-down

• CI results artifact: homeboy-ci-results-homeboy-review-audit-homeboy-Linux contains immediate command JSON for this action invocation.
• Observation artifact: homeboy-observations-homeboy-review-audit-homeboy-Linux contains exported Homeboy run history for deeper queries.
• Drill-down: download the observation artifact, then run homeboy runs import <dir>, homeboy runs list, and homeboy runs findings <run-id>.
• Artifacts are attached to the workflow run: https://github.com/Extra-Chill/homeboy/actions/runs/29223505287

Tooling versions

• Homeboy CLI: homeboy 0.283.1+bab65bc99e1d-dirty+bab65bc99
• Extension: rust from https://github.com/Extra-Chill/homeboy-extensions
• Extension revision: f80cdf4c
• Action: unknown@unknown