I had the following code:

void writeTask() {
   slowPart();
   foreach(subtask) {
       writeTask();
   }
}

after profiling the code it was not clear which method in slowPart() was slow
because it was called from different recursion levels. 

I reorganized the code like this:

void writeTask() {
   writeTask0();
   foreach(subtask) {
       writeTask();
   }
}

void writeTask0() {
   slowPart();
}

and profiled only writeTask0() and was able to find the slow method. Please add
the possibility to "collapse" recursive calls and show sums of spent times.