"rasa test core --evaluate-model-directory" should have directory as default

[wochinge]

Steps to Reproduce:

1. Hava a rasa project (e.g. by doing rasa init)
2. Train a single model (rasa train)
3. Run rasa test core --evaluate-model-directory

Problem:

By default rasa test core has the latest model in the ./models directory as default value for the --model parameter.

This can lead to the following error:
In case there is a regular model in the models directory, this will be assumed as latest model. However, rasa test core --evaluate-model-directory expects a directory and will fail with an error like this:

Traceback (most recent call last):
  File "/Users/tobias/.pyenv/versions/3.7/bin/rasa", line 11, in <module>
    load_entry_point('rasa', 'console_scripts', 'rasa')()
  File "/Users/tobias/Workspace/stack/rasa/__main__.py", line 76, in main
    cmdline_arguments.func(cmdline_arguments)
  File "/Users/tobias/Workspace/stack/rasa/cli/test.py", line 83, in test_core
    test_core_models_in_directory(model_path, stories, output)
  File "/Users/tobias/Workspace/stack/rasa/test.py", line 26, in test_core_models_in_directory
    number_of_stories = io_utils.read_json_file(story_n_path)
  File "/Users/tobias/Workspace/stack/rasa/utils/io.py", line 144, in read_json_file
    content = read_file(filename)
  File "/Users/tobias/Workspace/stack/rasa/utils/io.py", line 136, in read_file
    with open(filename, encoding=encoding) as f:
NotADirectoryError: [Errno 20] Not a directory: 'models/20191204-102617.tar.gz/num_stories.json'

Proposed Solutions

This can by automatically taking the directory instead of the file in case the user used the --evaluate-model-directory flag.

Other things to check and potentially create a new issue for it:
In case the models directory is empty (except the run folders from the comparison training) the default value for --models is None. So we have to actively check if there is a None value in the passed models list.