sentencizer should handle empty docs

Seen in https://github.com/jjlatval/cse3

```
root_indexer@68372[I]: (raised from /usr/local/lib/python3.7/site-packages/jina/importer.py:120)
segmenter@68168[I]:msg = self._post_hook(self._handle(self._pre_hook(msg)))
root_indexer@68372[I]:UserWarning: Install `uvloop` via `pip install "jina[uvloop]"` for better performance. (raised from /usr/local/lib/python3.7/site-packages/jina/helper.py:883)
segmenter@68168[I]:File "/usr/local/lib/python3.7/site-packages/jina/peapods/runtimes/zmq/zed.py", line 203, in _handle
segmenter@68168[I]:peapod_name=self.name,
segmenter@68168[I]:File "/usr/local/lib/python3.7/site-packages/jina/peapods/runtimes/request_handlers/data_request_handler.py", line 162, in handle
segmenter@68168[I]:field='groundtruths',
segmenter@68168[I]:File "/usr/local/lib/python3.7/site-packages/jina/executors/__init__.py", line 190, in __call__
segmenter@68168[I]:self, **kwargs
segmenter@68168[I]:File "/usr/local/lib/python3.7/site-packages/jina/executors/decorators.py", line 103, in arg_wrapper
segmenter@68168[I]:return fn(*args, **kwargs)
segmenter@68168[I]:File "/workspace/sentencizer.py", line 63, in segment
segmenter@68168[I]:flat_docs = docs.traverse_flat(traversal_path)
```

Start with the reproduction linked in the issue and inspect sentencizer.py at line 63, especially the segment method and its call to docs.traverse_flat(traversal_path). Reproduce the empty-document case and make the sentencizer complete without the reported failure; no test file is named in the issue.

Written by the indexing model from the issue text.