Make readability-function-cognitive-complexity threshold more strict (#365)

Functions with scores beyond 15 are typically flagged as potentially
problematic. Empirical evidence suggests that such functions are more
likely to correlate with higher error rates or maintenance challenges.
It becomes increasingly hard for developers to follow the control flow,
and these functions are prime candidates for refactoring into simpler,
more modular parts.

Author: aobolensk

Diff for: .clang-tidy

@@ -85,7 +85,8 @@ CheckOptions:
     value:           lower_case
   - key:             readability-identifier-naming.IgnoreMainLikeFunctions
     value:           1
+  # Functions with scores beyond 15 are typically flagged as potentially problematic (empirically)
   - key:             readability-function-cognitive-complexity.Threshold
-    value:           25  # default: 25
+    value:           15  # default: 25
   - key: misc-include-cleaner.IgnoreHeaders
     value: '(opencv2/.*|__chrono/.*)'

Diff for: tasks/mpi/example/src/ops_mpi.cpp

@@ -27,25 +27,22 @@ bool nesterov_a_test_task_mpi::TestTaskMPI::ValidationImpl() {
 bool nesterov_a_test_task_mpi::TestTaskMPI::RunImpl() {
   int rank = -1;
   MPI_Comm_rank(MPI_COMM_WORLD, &rank);
-  if (rank == 0) {
-    // Multiply matrices
+
+  auto multiply = [this](bool row_major) {
     for (int i = 0; i < rc_size_; ++i) {
       for (int j = 0; j < rc_size_; ++j) {
+        int sum = 0;
         for (int k = 0; k < rc_size_; ++k) {
-          output_[(i * rc_size_) + j] += input_[(i * rc_size_) + k] * input_[(k * rc_size_) + j];
-        }
-      }
-    }
-  } else {
-    // Multiply matrices
-    for (int j = 0; j < rc_size_; ++j) {
-      for (int k = 0; k < rc_size_; ++k) {
-        for (int i = 0; i < rc_size_; ++i) {
-          output_[(i * rc_size_) + j] += input_[(i * rc_size_) + k] * input_[(k * rc_size_) + j];
+          int a = input_[(row_major ? i : k) * rc_size_ + (row_major ? k : i)];
+          int b = input_[(row_major ? k : j) * rc_size_ + (row_major ? j : k)];
+          sum += a * b;
         }
+        output_[(i * rc_size_) + j] += sum;
       }
     }
-  }
+  };
+
+  multiply(rank == 0);
   MPI_Barrier(MPI_COMM_WORLD);
   return true;
 }