[Perf] OpenCL vs CUDA performance for Black Hole Example

Description
===========
The performance of the OpenCL back end is much worse than the CUDA back end on the same GPU when running the Black Hole example. May be worth investigating.

System Information
------------------
ArrayFire Version: main
Device: Nvidia RTX 4060 Laptop GPU
Operating System: Ubuntu 24.04
Driver version: 570.124.06
