Make mockable backend for IO

[Zluurk]

For all IO:

• Http request
• Ftp request
• File
• MLLP
• Futures IO
• ...

We should create a unified backend that can be used by user defined nodes and allow intelligent mock.

With such backend we can easily test a channel without mock the whole node and be sure that all data are correctly read from node.

For example with http request backend with should be able to specify a response when we have a specific request.

Another example: file backend should be memory only to avoid writing file by mistake.

For file backend we can imagine something like that:

class FileBackend():
    async def open(self, path, mode):
        if in_test_mode():
            # TODO should verify file don't previously exists
            # or mock is defined especially for this path
            mocked_file = StringIO()
            self.open_files.append({
                'path': path,
                'mode': mode,
                'file': mocked_file
            )
            return mocked_file
        else:
            return open(path, mode)

This mock should be used for every file modification in pypeman.