fix(mmdet/base_dense_head): 🐛 correct slicing issue in score softmax for single-class models (open-mmlab#2827)

The softmax function previously sliced out the background class, which
caused an IndexError in single-class model export. This update removes
the slicing of background class scores, fixing the issue when exporting
RetinaNet with single-class configurations.

- Updated `base_dense_head__predict_by_feat` to prevent IndexError when
handling single-class outputs.
- This change is particularly relevant for RetinaNet model exports.

Author: chuzhdontcode

mmdeploy/codebase/mmdet/models/dense_heads/base_dense_head.py

@@ -113,7 +113,7 @@ def base_dense_head__predict_by_feat(
         if self.use_sigmoid_cls:
             scores = scores.sigmoid()
         else:
-            scores = scores.softmax(-1)[:, :, :-1]
+            scores = scores.softmax(-1)
         if with_score_factors:
             score_factors = score_factors.permute(0, 2, 3,
                                                   1).reshape(batch_size,