Dispose/Close/Finish does not wait trace to get flushed.

Describe the bug When calling on of the methods Tracer.Dispose or Tracer.Close or Span.Finish the expected behavior is data to be sent to agent. Span.Finish says in its documentation: “Record the end time of the span and flushes it to the backend. After the span has been finished all modifications will be ignored”.

To Reproduce Steps to reproduce the behavior:

1. Create simple .NET Console application
2. Install Datadog.Trace NUGET version 1.18
3. Write code to trace a span and don’t delay application closing and run it:

        public static void Main(string[] args)
        {
            using (Scope scope = Tracer.Instance.StartActive("manual.fastshutdown"))
            {
                scope.Span.Type = SpanTypes.Custom;
                
            }
            
            //Thread.Sleep(10000); //uncomment this line to get trace into datadog
        }

1. Check datadog APM traces: there is no manual.fastshutdown trace in it.
2. uncomment commented line and run again
3. Check that trace manual.fastshutdown appeared in datadog

Runtime environment (please complete the following information):

• Instrumentation mode: NuGet package
• Tracer version: 1.18
• OS: Windows
• CLR: .NET Framework 4.8

[kehigginbotham]

I have discovered a viable work-around for my use case.

Traces are flushed if the flow of control exits the Scope’s scope gracefully. In the event of an ApplicationExit or other event initiating shutdown, the active Scope is not flushed. By adding the (VB.NET) code below in my ApplicationExit handler, I am able to force the active Scope to flush.

If Tracer.Instance.ActiveScope IsNot Nothing
    Tracer.Instance.ActiveScope.Dispose()
End If

[kevingosse]

I am closing this issue.

For future reference:

• When the process exits gracefully, the tracer should be able to flush the pending traces. If not, this is the bug, please open a issue to report it
• When the process is about to exit brutally (threads being aborted for instance), you can first call Tracer.Instance.ForceFlushAsync() to avoid losing the pending traces