Merge pull request #1815 from RosettaCommons/roccomoretti/fix_integration_unicode
Adjust how unicode encode/decode works on benchmark integration test.
A previous fix got the order of encode/decode wrong. This commit changes how the benchmark server tests handle unicode. Specifically, I'm attempting to use the recommended decode/encode at input/output approach. That is, read in files in a Unicode (UTF-8) aware manner, deal with strings internally as Unicode, and then output in a Unicode (UTF-8) aware manner.
This should fix the big failure of the integration.tutorials test that's currently happening on master.