supersize: Don't cluster by default. Make Diff() accept only SizeInfo.

tools/binary_size/libsupersize/file_format.py

Index: tools/binary_size/libsupersize/file_format.py
diff --git a/tools/binary_size/libsupersize/file_format.py b/tools/binary_size/libsupersize/file_format.py
index 867d29c5e063030a817baa5eabe231381ff8286a..e58494daacc7e557bc587308824dce87fa82fcf5 100644
--- a/tools/binary_size/libsupersize/file_format.py
+++ b/tools/binary_size/libsupersize/file_format.py
@@ -45,7 +45,7 @@ def _SaveSizeInfoToFile(size_info, file_obj):
   # Store a single copy of all paths and have them referenced by index.
   # Using an OrderedDict makes the indices more repetitive (better compression).
   path_tuples = collections.OrderedDict.fromkeys(
-      (s.object_path, s.source_path) for s in size_info.raw_symbols)
+      (s.object_path, s.source_path) for s in size_info.symbols)
   for i, key in enumerate(path_tuples):
     path_tuples[key] = i
   file_obj.write('%d\n' % len(path_tuples))
@@ -53,7 +53,7 @@ def _SaveSizeInfoToFile(size_info, file_obj):
   _LogSize(file_obj, 'paths')  # For libchrome, adds 200kb.
 
   # Symbol counts by section.
-  by_section = models.SymbolGroup(size_info.raw_symbols)
+  by_section = models.SymbolGroup(size_info.symbols)
   by_section = by_section.GroupBySectionName().SortedByName()
   file_obj.write('%s\n' % '\t'.join(g.name for g in by_section))
   file_obj.write('%s\n' % '\t'.join(str(len(g)) for g in by_section))
@@ -181,7 +181,8 @@ def _LoadSizeInfoFromFile(file_obj):
       raw_symbols[symbol_idx] = new_sym
       symbol_idx += 1
 
-  return models.SizeInfo(section_sizes, raw_symbols, metadata=metadata)
+  return models.SizeInfo(section_sizes, models.SymbolGroup(raw_symbols),
+                         metadata=metadata)
 
 
 def SaveSizeInfo(size_info, path):