Adding Story Cloze which requires manual download

[zaidalyafeai]

To allow manual download of a dataset, we can add a data_dir field in the builder. It will be helpful to know what the conf variable does.

def get_dataset_builder(path, conf=None):
    "Get a dataset builder from name and conf."
    module_path = datasets.load.prepare_module(path, dataset=True)
    builder_cls = datasets.load.import_main_class(module_path[0], dataset=True)
    if conf:
        builder_instance = builder_cls(name=conf, cache_dir=None, hash=module_path[1],
        data_dir = 'path/to/local/')
    else:
        builder_instance = builder_cls(cache_dir=None, hash=module_path[1])
    return builder_instance

Also, since Story Cloze is a special case we can add it to the dataset_list field i.e dataset_list += ["story_cloze"].
A couple of things that we need to figure out ?

1. What is the directory for storing the dataset locally in the server?
2. Since the dataset is not public we should only show a few samples to help in making the templates.
3. If the PR is not merged by the weekend we should consider adding custom data loaders.

[zaidalyafeai]

being resolved here #443

[zaidalyafeai]

Other issues also solved in #447