refactor: replace scatter_ with argsort+index_select for better cross-platform compatibility

- Replace tensor.scatter_() with torch.argsort() + torch.index_select(out=)
- Uses fundamental PyTorch ops for improved reliability across CUDA/ROCm/MPS
- Aggressive early tensor deletion to minimize memory overhead
- Affects _histogram_matching_channel and _histogram_match_1d in color_fix.py

Related: #351

src/utils/color_fix.py

@@ -490,6 +490,7 @@ def _histogram_matching_channel(source: Tensor, reference: Tensor, device: torch
     # Sort both arrays
     source_sorted, source_indices = torch.sort(source_flat)
     reference_sorted, _ = torch.sort(reference_flat)
+    del reference_flat
     
     # Quantile mapping
     n_source = len(source_sorted)
@@ -503,14 +504,17 @@ def _histogram_matching_channel(source: Tensor, reference: Tensor, device: torch
         ref_indices = (source_quantiles * (n_reference - 1)).long()
         ref_indices.clamp_(0, n_reference - 1)
         matched_sorted = reference_sorted[ref_indices]
-        del source_quantiles, ref_indices
+        del source_quantiles, ref_indices, reference_sorted
     
-    # Reconstruct with matched values
-    matched_flat = torch.empty_like(source_flat)
-    matched_flat.scatter_(0, source_indices, matched_sorted)
-    del source_flat, reference_flat, source_sorted, source_indices, reference_sorted, matched_sorted
+    del source_sorted
     
-    return matched_flat.reshape(original_shape)
+    # Reconstruct using index_select with out param (portable, memory-efficient)
+    inverse_indices = torch.argsort(source_indices)
+    del source_indices
+    torch.index_select(matched_sorted, 0, inverse_indices, out=source_flat)
+    del matched_sorted, inverse_indices
+    
+    return source_flat.reshape(original_shape)
 
 
 def hsv_saturation_histogram_match(content_feat: Tensor, style_feat: Tensor, debug: Optional['Debug'] = None) -> Tensor:
@@ -748,11 +752,16 @@ def _histogram_match_1d(source: Tensor, reference: Tensor, device: torch.device)
         ref_indices = (source_quantiles * (n_reference - 1)).long()
         ref_indices.clamp_(0, n_reference - 1)
         matched_sorted = reference_sorted[ref_indices]
-        del source_quantiles, ref_indices
+        del source_quantiles, ref_indices, reference_sorted
     
+    del source_sorted
+    
+    # Reconstruct using index_select with out param (portable, memory-efficient)
     matched = torch.empty_like(source)
-    matched.scatter_(0, source_indices, matched_sorted)
-    del source_sorted, source_indices, reference_sorted, matched_sorted
+    inverse_indices = torch.argsort(source_indices)
+    del source_indices
+    torch.index_select(matched_sorted, 0, inverse_indices, out=matched)
+    del matched_sorted, inverse_indices
     
     return matched