Improve setup.py so that it is more sophisticated

[mcflyhalf]

The setup basically comes up with the Item similarity matrix.

Improve the setup.py file to be able to come up with the files required to run the recommender with its only input as a text-based historical file

Text-based historical record file

In this case, the input is a file containing plain text in csv format. The system will take this file and use it to generate an item similarity matrix. A few things which will also need to be generated along the way are:
* Unique item list. A list of each unique item appearing in the input file
* Unique IDs for each unique item. This may be the name as given in the unique item list or a numerical UID or some other form of appropriate unique identifier
* The item similarity matrix in some form such as a database table.