BUG: Let elastix SplitRequestedRegion override ITK's ImageSource

Adjusted `ParabolicErodeDilateImageFilter::SplitRequestedRegion` to ensure that its signature matches `ImageSource::SplitRequestedRegion` (again). Following ITK commit InsightSoftwareConsortium/ITK@c701410, June 14, 2011, which modified the signature of `ImageSource::SplitRequestedRegion`, by replacing signed with unsigned integers.

Similar to pull request #1118 commit 34bdc3b "BUG: Let `elx::ConjugateGradientFRPR` override ITK's FRPROptimizer"

Note: This commit changes the behavior (restoring the originally intended behavior), because `ParabolicErodeDilateImageFilter::SplitRequestedRegion` does avoid the current dimension (`m_CurrentDimension`), when setting the `splitAxis`.

Bug found by macos-12/clang `-Woverloaded-virtual` warnings.

Common/itkParabolicErodeDilateImageFilter.h

@@ -142,8 +142,8 @@ class ITK_TEMPLATE_EXPORT ParabolicErodeDilateImageFilter : public ImageToImageF
   void
   GenerateData() override;
 
-  int
-  SplitRequestedRegion(int i, int num, OutputImageRegionType & splitRegion);
+  unsigned int
+  SplitRequestedRegion(unsigned int i, unsigned int num, OutputImageRegionType & splitRegion) override;
 
   void
   ThreadedGenerateData(const OutputImageRegionType & outputRegionForThread, ThreadIdType threadId) override;

Common/itkParabolicErodeDilateImageFilter.hxx

@@ -60,10 +60,10 @@ ParabolicErodeDilateImageFilter<TInputImage, doDilate, TOutputImage>::ParabolicE
 
 
 template <typename TInputImage, bool doDilate, typename TOutputImage>
-int
+unsigned int
 ParabolicErodeDilateImageFilter<TInputImage, doDilate, TOutputImage>::SplitRequestedRegion(
-  int                     i,
-  int                     num,
+  unsigned int            i,
+  unsigned int            num,
   OutputImageRegionType & splitRegion)
 {
   // Get the output pointer
@@ -98,7 +98,7 @@ ParabolicErodeDilateImageFilter<TInputImage, doDilate, TOutputImage>::SplitReque
   int                                            maxThreadIdUsed = (int)::ceil(range / (double)valuesPerThread) - 1;
 
   // Split the region
-  if (i < maxThreadIdUsed)
+  if (static_cast<intmax_t>(i) < maxThreadIdUsed)
   {
     splitIndex[splitAxis] += i * valuesPerThread;
     splitSize[splitAxis] = valuesPerThread;