Refactor mo_task_spawn() for the new scheduler

Move sched_enqueue_task() into a critical section to protect ready_queue[]
integrity, as the API modifies shared scheduler resources.

Initialize the embedded rq_node when a task spawns and set its next pointer
to NULL to ensure deterministic linkage for ready-queue insertion.

Bind the initial task slot using rq_node instead of the global task list,
matching the new ready-queue selection model.

This aligns task-spawn behavior with rq_node-based scheduling semantics.

Author: vicLin8712

kernel/task.c

@@ -820,8 +820,15 @@ int32_t mo_task_spawn(void *task_entry, uint16_t stack_size_req)
     tcb->id = kcb->next_tid++;
     kcb->task_count++; /* Cached count of active tasks for quick access */
 
+    /* Binding ready queue node */
+    tcb->rq_node.data = tcb;
+    tcb->rq_node.next = NULL;
+
     if (!kcb->task_current)
-        kcb->task_current = node;
+        kcb->task_current = &tcb->rq_node;
+
+    /* Push node to ready queue */
+    sched_enqueue_task(tcb);
 
     CRITICAL_LEAVE();
 
@@ -841,7 +848,6 @@ int32_t mo_task_spawn(void *task_entry, uint16_t stack_size_req)
 
     /* Add to cache and mark ready */
     cache_task(tcb->id, tcb);
-    sched_enqueue_task(tcb);
 
     return tcb->id;
 }