Is your feature request related to a problem? Please describe.
Now this code is available only for me and @snk4tr as part of another project. Propose to simplify it and add as a standalone script into the library, so that everyone could verify metric performance.

Describe the solution you'd like
Add benchmark.py file into tests folder with support for TID2013 and KADID10k databases evaluation.
Create README section with results.

Additional context
I've run all the tests locally and it seems that not all implementations from our library are now consistent with results from papers. Here is table for TID2013 database