Before, Unicode characters would not decode properly. Now, if a replacement character of '\ufffd' is found in the decoded output, it will continue requesting more completion tokens until it is able to decode without a replacement character / decode error.
Fix for #30
Before, Unicode characters would not decode properly. Now, if a replacement character of '\ufffd' is found in the decoded output, it will continue requesting more completion tokens until it is able to decode without a replacement character / decode error.