[core] Fix TThreadExecutor's MapReduce on vectors with chunks

The last chunk of partial results of MapReduce with std::vectors had the
size of all other chunks. This is dangerous, since the default vector
values might cause unexpected results of the Reduce function.

Test added to verify the correctness.

Author: ikabadzhov

core/imt/inc/ROOT/TThreadExecutor.hxx

@@ -412,8 +412,8 @@ namespace ROOT {
       using retType = decltype(func(args.front()));
       std::vector<retType> reslist(actualChunks);
       auto lambda = [&](unsigned int i) {
-         std::vector<retType> partialResults(step);
-         for (unsigned j = 0; j < step && (i + j) < nToProcess; j++) {
+         std::vector<retType> partialResults(std::min(step, nToProcess - i));
+         for (unsigned j = 0; j < partialResults.size(); j++) {
             partialResults[j] = func(args[i + j]);
          }
          reslist[i / step] = Reduce(partialResults, redfunc);
@@ -444,8 +444,8 @@ namespace ROOT {
       using retType = decltype(func(args.front()));
       std::vector<retType> reslist(actualChunks);
       auto lambda = [&](unsigned int i) {
-         std::vector<retType> partialResults(step);
-         for (unsigned j = 0; j < step && (i + j) < nToProcess; j++) {
+         std::vector<retType> partialResults(std::min(step, nToProcess - i));
+         for (unsigned j = 0; j < partialResults.size(); j++) {
             partialResults[j] = func(args[i + j]);
          }
          reslist[i / step] = Reduce(partialResults, redfunc);

core/imt/test/testRTaskArena.cxx

@@ -246,4 +246,25 @@ TEST(TThreadExecutor, ThreadSafety) {
    EXPECT_TRUE(std::equal(counters.begin(), counters.end(), target.begin()));
 }
 
+// Checking if we correctly handle uneven chunks
+TEST(TThreadExecutor, StdVectorChunks)
+{
+   ROOT::TThreadExecutor ttex;
+   auto func = [](int x) -> int { return x; };
+   // redfunc must be such that does not have 0 as identity (i.e. not addition but multiplication)
+   auto redfunc = [](const std::vector<int> &v) {
+      return std::accumulate(v.begin(), v.end(), 1, std::multiplies<int>());
+   };
+
+   // will be calculating 7 factorial = 5040, const and non-const vectors to invoke different overloads
+   std::vector<int> vec{1, 2, 3, 4, 5, 6, 7};
+   const std::vector<int> cvec{1, 2, 3, 4, 5, 6, 7};
+
+   EXPECT_EQ(ttex.MapReduce(func, vec, redfunc, 3), 5040); // with 3 chunks, last chunk is smaller
+   EXPECT_EQ(ttex.MapReduce(func, cvec, redfunc, 3), 5040);
+
+   EXPECT_EQ(ttex.MapReduce(func, vec, redfunc, 9), 5040); // with 9 chunks, 2 empty chunks
+   EXPECT_EQ(ttex.MapReduce(func, cvec, redfunc, 9), 5040);
+}
+
 #endif