torch.cuda.get_device_capability() in gen_trtllm_comm_module() breaks AOT build

[tiran]

In PR #1249 you added code that uses at torch.cuda.get_device_capability(). This break ahead of time compilation. get_device_capability() only works on machines with CUDA hardware. On build machines without CUDA hardware, the function fails:

RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx

[yzh119]

Good catch and yes we should fix the behavior.