[llvm] Modify dwarf verification JSON to include detailed counts by sub-category (PR #128018)

[via llvm-commits]

================
@@ -30,18 +30,28 @@ class DWARFDebugAbbrev;
 class DataExtractor;
 struct DWARFSection;
 
+struct AggregationData {
+  unsigned OverallCount;
+  std::map<std::string, unsigned> DetailedCounts;
----------------
youngd007 wrote:

What would be your alternative suggestion?  The counts are logged one at a time, so the sub-keys can come in out of order (probably, right @clayborg ?).  So we need to update a count for one key.  Otherwise, we would, what store a list of pairs?  Iterate the list to find pair with the right key and increment the count there?  Is that more performant in some way?  I don't imagine we have hundreds of sub-keys.  maybe tens at worse case.

https://github.com/llvm/llvm-project/pull/128018