add evaluation.py to ml branch

[OlegMatveevS]

Raw version of the evaluation.py script

#233

[iburakov]

Oh, look at new ModelsStore too. Can be reused here, I guess, to generalize model loading logic. It'll have to be moved to components from mlbackend then.

[iburakov]

getting there! relatively minor fixes still required, but during next pass I'll probably execute/test it myself and implement all final needed refactorings.

also try this (leave it up to me if it seems complex):

• make example path a main argument
• refactor CLI argument parsing to do so via argparse module (see other scripts for examples)
• implement automatic test doing a simple main(...) call on a quick example to make sure this script doesn't break over time

[ryukzak]

We need:

• rebase on the master
• format code via black, isort and check by flake8 (see students chat for details)
• we need by script arguments:
  • list of examples (not path). E.g.: evaluation.py example/*.lua, or evaluation.py example/pid.lua example/sum.lua
  • select evaluation methods (should work ML synthesis)
  • pass nitta arguments.
• Add usage example (with output) to readme.md (for master branch -- without ML, for ml branch -- with).