[SC] Limitar el número de tareas en ejecución dentro de un TaskGroup

Swift's cooperative thread pool does not prevent resource exhaustion when many tasks (e.g., 1000 image downloads) are scheduled simultaneously via TaskGroup.addTask. The sliding window pattern solves this by seeding a fixed number of initial tasks and using group.next() to wait for one to finish before launching the next, keeping concurrency capped. ArraySlice with popFirst() serves as an efficient pending-work queue that avoids copying the original array.