[Enhance] Add a comment for the use of masked_conv on Ascend device (#2689)

* Update masked_conv.py

* Update mmcv/ops/masked_conv.py

---------
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

mmcv/ops/masked_conv.py

@@ -46,6 +46,8 @@ class MaskedConv2dFunction(Function):
         out_channel, in_channel, kernel_h, kernel_w = weight.size()
 
         if features.device.type == 'npu':
+            # For the NPU implementation of the masked_conv operator, it is
+            # not recommended to use the scenario where bias=false.
             import torch_npu
             output = torch_npu.npu_conv2d(
                 features,