GPU Metal: create the device events

The event slots were only placement-constructed as nil ids, so the first
StreamWaitForEvents threw "object cannot be nil" from encodeWaitForEvent.
Create an MTLSharedEvent per slot, as the CUDA backend creates its
events, and release them on exit.

Author: ktf

GPU/GPUTracking/Base/metal/GPUReconstructionMetal.mm

@@ -137,7 +137,12 @@
 
   for (uint32_t i = 0; i < mEvents.size(); i++) {
     auto* events = (id<MTLSharedEvent>*)mEvents[i].data();
-    new (events) id<MTLSharedEvent>[ mEvents[i].size() ];
+    for (uint32_t j = 0; j < mEvents[i].size(); j++) {
+      events[j] = [mInternals->device newSharedEvent];
+      if (events[j] == nil) {
+        GPUErrorReturn("Error creating Metal event");
+      }
+    }
   }
 
   return (0);
@@ -148,6 +153,13 @@
   // Tear down Metal
   SynchronizeGPU();
 
+  for (uint32_t i = 0; i < mEvents.size(); i++) {
+    auto* events = (id<MTLSharedEvent>*)mEvents[i].data();
+    for (uint32_t j = 0; j < mEvents[i].size(); j++) {
+      [events[j] release];
+    }
+  }
+
   if (mMaster == nullptr) {
     if (mDeviceMemoryBase) {
       [mInternals->mem_gpu release];