Issue #13624: Write a specialized UTF-8 encoder to allow more optimization

The main bottleneck was the PyUnicode_READ() macro.
3 files changed