Display origin of template instantiations

[vittorioromeo]

Addresses #93. Could be improved in the future, but it's already quite useful. Example output:

**** Templates that took longest to instantiate:
   602 ms: std::vformat_to<std::__format::_Sink_iter<char>> (5 times, avg 120 ms)
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/StackTrace.cpp.obj
       1 time(s): .//_deps/cpptrace-build/CMakeFiles/cpptrace-lib.dir/Unity/unity_0_cxx.cxx.obj
       1 time(s): .//test/CMakeFiles/test-sfml-system.dir/System/Time.test.cpp.obj
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/Clock.cpp.obj
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/ThreadPool.cpp.obj
   601 ms: std::__format::__do_vformat_to<std::__format::_Sink_iter<char>, char... (5 times, avg 120 ms)
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/StackTrace.cpp.obj
       1 time(s): .//_deps/cpptrace-build/CMakeFiles/cpptrace-lib.dir/Unity/unity_0_cxx.cxx.obj
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/ThreadPool.cpp.obj
       1 time(s): .//test/CMakeFiles/test-sfml-system.dir/System/Time.test.cpp.obj
       1 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/Clock.cpp.obj

**** Template sets that took longest to instantiate:
   717 ms: std::formatter<$>::format<$> (117 times, avg 6 ms)
      23 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/StackTrace.cpp.obj
      23 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/__/Base/ThreadPool.cpp.obj
      24 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/Clock.cpp.obj
      24 time(s): .//test/CMakeFiles/test-sfml-system.dir/System/Time.test.cpp.obj
      23 time(s): .//_deps/cpptrace-build/CMakeFiles/cpptrace-lib.dir/Unity/unity_0_cxx.cxx.obj
   676 ms: std::common_reference<$> (881 times, avg 0 ms)
       1 time(s): .//test/CMakeFiles/test-sfml-audio.dir/Audio/SoundBuffer.test.cpp.obj
       3 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/FileUtils.cpp.obj
       1 time(s): .//test/CMakeFiles/test-sfml-audio.dir/Audio/InputSoundFile.test.cpp.obj
       6 time(s): .//src/SFML/Network/CMakeFiles/sfml-network.dir/Http.cpp.obj
       7 time(s): .//test/CMakeFiles/test-sfml-network.dir/Network/Packet.test.cpp.obj
       6 time(s): .//test/CMakeFiles/test-sfml-graphics.dir/Graphics/Transform.test.cpp.obj
      17 time(s): .//src/SFML/Window/CMakeFiles/sfml-window.dir/SDLLayer.cpp.obj
      30 time(s): .//src/SFML/System/CMakeFiles/sfml-system.dir/Clock.cpp.obj

This PR needs work, but I'd like to get some feedback first on how to proceed. It's mostly a hack at this point.

[vittorioromeo]

Some questions for @aras-p

[vittorioromeo]

This will be removed, was just for testing.

[vittorioromeo]

I don't like this approach, but I need guidance on where these things should go.

[vittorioromeo]

Is there a way we can avoid adding the extra field?

[vittorioromeo]

Can we get more precise information other than the event filename, somehow?

[leekillough]

FYI, for curiosity, I tried this, but I am getting data corruption causing an infinite loop.

src/BuildEvents.cpp

701             r.Read(childCount);
702             e.children.resize(childCount);

(gdb) p childCount
$7 = 21474836480

The attached capture file was used as input. It works on main, but hangs after data corruption on this branch.

My executable used to capture was out of sync with my executable used for analysis.

[vittorioromeo]

@aras-p: any interest in getting this PR improved and merged?