Make evaluation.py more informative

[xavierfigueroav]

The evaluation script should

• Allow to print a comparison table with all the algorithms supported by atarashi. You can find examples of comparison tables in Improve TF-IDF agent by tuning matches threshold #95 and [Proposal] Improve the speed of matching  #65.
• Allow to print a confusion matrix so that we can easily do error analysis to make decisions on how to improve current agents or implement new ones.

[prasanna7codes]

i have made changes in the evaluator.py and tfidf.py , but I am not able to raise a pull request , please let me know how I can deliver it to you