Merge pull request #2545 from RosettaCommons/roccomoretti/utf_decode_replace
Be more robust with UTF parsing.
Add errors=replace to the UTF decode statements in Python scripts. Instead of crashing when input contains non-parsable data, replace it with the codepoint-missing character.
Doesn't completely solve the issue, as Python2 will still crash when redirecting Unicode-containing characters. -- That's not really something we can fix from within the script, though. You need to set the `PYTHONIOENCODING` environment variable in the launching shell to `utf_8` or equivalent.