Improvements when indexing through Applications

[davidmezzetti]

Make the following improvements to help indexing through applications.

• In Application.add, cast the input documents batch to a list. This will automatically convert a generator if that's provided. Note that this method is designed to work with small batches, not the entire dataset

• Default graph and topics configuration if it's None. This is the same logic already available in scoring. This makes it possible to have cleaner input YAML syntax.