Add "--expected-fail" option to run_framework.py

[Hardcode84]

This option allows to specify list of benchmarks which are known to fail so entire run will still be considered successful. Also, it can detect if previously failed tests unexpectedly passed.