Use @autoreleasepool to avoid memory leaks

There appears to be a known memory leak when using the
`MLTCommandBuffer`. It is suggested to use `@autoreleasepool` in
[1,2]

[1] https://developer.apple.com/forums/thread/662721
[2] https://forums.developer.apple.com/forums/thread/120931

This change-set wraps the `ggml_metal_graph_compute` in a
`@autoreleasepool`.

This commit addresses #5436

Author: irbull

ggml-metal.m

@@ -2577,7 +2577,9 @@ GGML_CALL static ggml_backend_buffer_type_t ggml_backend_metal_get_default_buffe
 GGML_CALL static bool ggml_backend_metal_graph_compute(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
     struct ggml_metal_context * metal_ctx = (struct ggml_metal_context *)backend->context;
 
-    return ggml_metal_graph_compute(metal_ctx, cgraph);
+    @autoreleasepool {
+        return ggml_metal_graph_compute(metal_ctx, cgraph);
+    }
 }
 
 GGML_CALL static bool ggml_backend_metal_supports_op(ggml_backend_t backend, const struct ggml_tensor * op) {