Now we are able to limit the number of spawned child processes

Now listening address, listening port and max_children count can be configured using the Settings module

Adding a cache for app module reducing seriously routing overhead

Regression may have been introduced about MONOSITE mode

- found a way to make it implements the socketserver.ForkinTCPServer features reducing the global code size (and removed a lot of dirty copy & paste from cpython sources)
- deleted useless imports

- updated scripts etc
- make tests pass

The server is listening, fork on new request, and then loads the context from URL.
Now we have to call properly the webui plugin in order to answer the request