The runtime now explicitly flushes trace data when a program crashes due to an uncaught panic. This means that more complete trace data will be available in a trace if the program crashes while tracing is active.