WIP: refactoring of the multiframe conversion implementation

This is an intermediate commit that aims to improve organization of the conversion
functionality, motivated by the various standing issues (e.g., QIICR#217, QIICR#192 and QIICR#150)
that are difficult to address with the current level of disarray in the code.

Some of the ideas are the following:

 - introduce basic hierarchy of classes to allow reuse of the common
   functionality between PM and SEG converters (MultiFrameConverter parent for
   ParametricMapConverter and SegmentationImageConverter)
 - do not use static functions, instead make better use of converter classes
 - factor out containers to keep information about ImageVolume and DICOMFrame
   to simplify conversion and book keeping
 - improve consistency of naming and code style

My plan to proceed with the refactoring is to implement "shadow" functions in the
updated classes while keeping the original functions initially, and then phase out
the original functions.

The current commit is just an intermediate milestone to open this for feedback
and keep development history at least somewhat incremental.

apps/seg/itkimage2segimage.cxx

@@ -91,7 +91,7 @@ int main(int argc, char *argv[])
     segmentations = segmentationsReordered;
   }
 
-  DcmDataset* result = dcmqi::ImageSEGConverter::itkimage2dcmSegmentation(dcmDatasets, segmentations, metadata, skipEmptySlices);
+  DcmDataset* result = dcmqi::SegmentationImageConverter::itkimage2dcmSegmentation(dcmDatasets, segmentations, metadata, skipEmptySlices);
 
   if (result == NULL){
     return EXIT_FAILURE;