[CUDA] Recycle CUDA events (#2604)

* Make CudaEvent a CudaHandle

* Add caching for CudaEvent

* Make sure cuda events are destroyed at last

* Fix headers

* SharedEvent => AtomicEvent

* RawCudaEvent => CudaEventHandle, CudaEventWrapper => CopyableCudaEvent

* Remove unneeded asserts

mlx/backend/cuda/worker.cpp

@@ -7,6 +7,7 @@ namespace mlx::core::cu {
 
 Worker::Worker()
     : signal_stream_(device(mlx::core::Device::gpu)),
+      signal_event_(cudaEventDisableTiming | cudaEventBlockingSync),
       worker_(&Worker::thread_fn, this) {}
 
 Worker::~Worker() {