fix(server,kernel): freestyle survives kill/restart at every phase (B4)

Three gaps found by live QA (kill between planning-complete and plan lock):
- launchSessionResumeWithRehydrate never handed off to FreestyleDriver, so a
  resumed freestyle session stranded at planning-complete with no plan lock
  and no phase 2; now locks+runs when planning completed and no
  ExecutionPlanLockedEvent exists yet.
- resume route 400'd on phase-2 sessions (workflowId freestyle-<sid> is
  compiled, never registered); now recompiles the locked plan after
  rehydrating the artifact cache.
- orchestrator.resume threw when currentStageId was terminal ('done') or
  unknown to the graph; now returns WorkflowResult.Completed.

core/kernel/src/main/kotlin/com/correx/core/kernel/orchestration/DefaultSessionOrchestrator.kt

@@ -219,6 +219,15 @@ class DefaultSessionOrchestrator(
     ): WorkflowResult {
         val stageId = orchestrationRepository.getState(sessionId).currentStageId
             ?: return WorkflowResult.Failed(sessionId, "resume: no currentStageId", retryExhausted = false)
+        if (!graph.stages.containsKey(stageId)) {
+            // Terminal pseudo-stage ("done") or a stage from another graph — nothing to
+            // re-enter. Treat as already completed instead of throwing inside enterStage.
+            log.info(
+                "[Orchestrator] resume: session={} stage={} not in graph '{}' — already terminal",
+                sessionId.value, stageId.value, graph.id,
+            )
+            return WorkflowResult.Completed(sessionId, stageId)
+        }
         log.info("[Orchestrator] resuming session={} workflow={} stage={}", sessionId.value, graph.id, stageId.value)
         val base = ExecutionContext(graph, sessionId, 0, stageId, config, null, null)
         val enriched = base.enrich()