Feature: PlanExecutor Coverage Boost
  Additional scenarios that exercise previously uncovered code paths
  in plan_executor.py, targeting lines 328-335, 384, 422, 458,
  598-599, 612, and 619.

  Background:
    Given the plan executor module is imported

  # --- _try_emit_metric success path (lines 328-330) ---

  Scenario: _try_emit_metric emits a metric when emitter is configured
    Given a PlanExecutor with a mock metrics emitter
    When I trigger metric emission with a valid key and value
    Then the metrics emitter should have received exactly one entry

  # --- _try_emit_metric exception path (lines 331-335) ---

  Scenario: _try_emit_metric silently handles emission failure
    Given a PlanExecutor with a failing metrics emitter
    When I trigger metric emission that causes an exception
    Then no exception should propagate from metric emission

  # --- _try_create_checkpoint returns None when sandbox is None (line 384) ---

  Scenario: _try_create_checkpoint returns None when no sandbox is resolvable
    Given a PlanExecutor with a checkpoint manager but no sandbox source
    When I attempt to create a plan executor checkpoint
    Then the checkpoint result should be None

  # --- _try_rollback_to_last_checkpoint returns False when sandbox is None (line 422) ---

  Scenario: _try_rollback_to_last_checkpoint returns False when no sandbox is resolvable
    Given a PlanExecutor with a checkpoint manager but no sandbox source
    When I attempt to rollback to the last checkpoint
    Then the executor rollback result should be False

  # --- _resolve_sandbox_for_checkpoint via execution_context (line 458) ---

  Scenario: _resolve_sandbox_for_checkpoint returns sandbox from execution context
    Given a PlanExecutor with an execution context that has a sandbox manager
    When I resolve a sandbox for checkpointing
    Then the resolved sandbox should come from the execution context

  # --- _enforce_guardrails raises PlanError on step limit (lines 598-599) ---

  Scenario: _enforce_guardrails raises PlanError when step limit is already reached
    Given a PlanExecutor with a guardrail service that blocks step limit
    When I enforce guardrails for a plan
    Then a PlanError about step limit should be raised

  # --- _enforce_guardrails_per_step returns when guardrails is None (line 612) ---

  Scenario: _enforce_guardrails_per_step returns early when guardrails is None
    Given a PlanExecutor with a guardrail service that returns no guardrails
    When I enforce per-step guardrails
    Then no exception should be raised from per-step enforcement

  # --- _enforce_guardrails_per_step raises PlanError on wall-clock (line 619) ---

  Scenario: _enforce_guardrails_per_step raises PlanError when wall-clock exceeded
    Given a PlanExecutor with a guardrail service that blocks wall-clock per step
    When I enforce per-step guardrails expecting wall-clock error
    Then a PlanError about wall-clock limit should be raised from per-step