Add support for non-blocking HTTP handlers

[rschlaikjer]

The current design of the HTTP server requires that the HTTP request finishes completely before any other sockets / their requests can be serviced.

For applications where processing an HTTP request may require a (slow) call to an external service, this means that the latency for concurrent calls stacks:

    def HandleHttpRequest(self, request: HttpRequest, **kwargs):
        time.sleep(3)
        return HttpResponse(request, statusCode=200)

root@sim:/root# time curl -I localhost:8091 & time curl -I localhost:8091
...
real	0m3.009s
real	0m6.012s

To improve this, add the option for request processors to indicate that the request is processing but not complete yet, allowing the framework to then spin other connections in parallel with the external work:

    def HandleHttpRequest(self, request: HttpRequest, **kwargs):
        if not request.userState:
            request.userState = time.time() + 3

        if time.time() < request.userState:
            raise HttpResponseProcessing

        return HttpResponse(request, statusCode=200)

root@sim:/root# time curl -I localhost:8091 & time curl -I localhost:8091
...
real	0m3.009s
real	0m3.010s