Enhanced CorrectionService.execute_revert to implement the full revert-mode
re-execution pipeline from the specification (§ Correction Flow, Revert Mode):
1. Resource rollback: When a CheckpointService is wired and a decision-aligned
checkpoint exists, delegates to rollback_to_checkpoint for real git reset.
Gracefully degrades when checkpointing is unavailable (returns False).
2. Reasoning rollback: Extracts actor_state_ref from the target decision's
context_snapshot and returns it in the result for downstream LangGraph
checkpoint restoration.
3. Guidance injection: Generates a user_intervention decision ID so callers
can record the correction guidance in the decision tree.
4. Phase transition: Signals that the plan should re-enter Strategize via
phase_transition_target="strategize" on the result.
Added four new fields to CorrectionResult: checkpoint_restored, actor_state_ref,
user_intervention_decision_id, and phase_transition_target — all with backward-
compatible defaults. Extended execute_correction dispatch to forward the new
decisions parameter.
Tests: 16 Behave BDD scenarios (features/revert_re_execution.feature) covering
checkpoint restoration, actor state recovery, phase transition, error paths,
subtree selectivity, and dispatch integration. 7 Robot Framework integration
tests (robot/revert_re_execution.robot). Coverage: 98% (above 97% threshold).
ISSUES CLOSED: #844
brent.edwards