- fix issues with training with external memory on cpu (#4487)

* - fix issues with training with external memory on cpu
   - use the batch size to determine the correct number of rows in a batch
   - use the right number of threads in omp parallalization if the batch size
     is less than the default omp max threads (applicable for the last batch)

* - handle scenarios where last batch size is < available number of threads
- augment tests such that we can test all scenarios (batch size <, >, = number of threads)

tests/cpp/common/test_column_matrix.cc

@@ -50,5 +50,23 @@ TEST(DenseColumnWithMissing, Test) {
   }
   delete dmat;
 }
+
+void
+TestGHistIndexMatrixCreation(size_t nthreads) {
+  /* This should create multiple sparse pages */
+  std::unique_ptr<DMatrix> dmat = CreateSparsePageDMatrix(1024, 1024);
+  omp_set_num_threads(nthreads);
+  GHistIndexMatrix gmat;
+  gmat.Init(dmat.get(), 256);
+}
+
+TEST(HistIndexCreationWithExternalMemory, Test) {
+  // Vary the number of threads to make sure that the last batch
+  // is distributed properly to the available number of threads
+  // in the thread pool
+  TestGHistIndexMatrixCreation(20);
+  TestGHistIndexMatrixCreation(30);
+  TestGHistIndexMatrixCreation(40);
+}
 }  // namespace common
 }  // namespace xgboost