fix: lifecycle bugs A+B (HasError guard, not_created hard error) + bump 5.0.72

provider/internal/resources_core/crud.go

@@ -153,14 +153,10 @@ func adoptExistingInstanceOnCreate(ctx context.Context, client *core.UniversalCl
 
 	switch state {
 	case StateNotCreated:
-		// P0.2: авто-cleanup orphan-инстанса при adopt_existing_on_create=true.
-		unlock := client.LockInstance(existing.InstanceUid)
-		defer unlock()
-		if err := client.RunInstanceOperationUniversal(ctx, existing.InstanceUid, "delete", nil); err != nil {
-			return "", fmt.Errorf("не удалось авто-удалить orphan-инстанс %s (статус: %s): %w. %s",
-				existing.InstanceUid, statusText, err, formatInstanceDetails(existing, serviceID, displayName))
-		}
-		return client.CreateGenericInstanceUniversalV6(ctx, serviceID, displayName, params)
+		// Философия (provider_philosophy.md раздел 8): not created → hard error,
+		// без auto-adopt/create. Пользователь должен проверить ресурс в ЛК.
+		return "", fmt.Errorf("инстанс %s в статусе 'not created' — проверьте ресурс в личном кабинете Nubes. %s",
+			existing.InstanceUid, formatInstanceDetails(existing, serviceID, displayName))
 
 	case StateRunning, StateRunningPending:
 		refIssues, err := ValidateRefParamsOnAdopt(ctx, client, serviceID, params, existing.InstanceUid)

provider/internal/resources_core/resource_diagnostics.go

@@ -61,7 +61,7 @@ func PlanExistingResourceDiagnostics(ctx context.Context, client *core.Universal
 	return diags
 }
 
-// CreateExistingResourceDiagnostics returns warnings for existing resources during create.
+// CreateExistingResourceDiagnostics returns diagnostics (errors for conflicts, warnings for adopt) during create.
 func CreateExistingResourceDiagnostics(ctx context.Context, client *core.UniversalClient, serviceID int, resourceName string, adoptExistingOnCreate bool) diag.Diagnostics {
 	return CreateExistingResourceDiagnosticsWithDomain(ctx, client, serviceID, resourceName, adoptExistingOnCreate, "")
 }