Update torch-ext/triton_kernels/target_info.py num_sms() to support pytorch-rocm cuda and gpt-oss models

torch-ext/triton_kernels/target_info.py

@@ -96,3 +96,5 @@ def num_sms():
         return torch.cuda.get_device_properties(0).multi_processor_count
     if is_xpu():
         return torch.xpu.get_device_properties(0).max_compute_units
+    if is_hip() and torch.cuda.is_available():
+        return torch.cuda.get_device_properties(0).multi_processor_count