Rewrite UTF-8 validation in shift-based DFA

This gives plenty of performance increase on validating strings with
many non-ASCII codepoints, which is the normal case for almost every
non-English content.

Shift-based DFA algorithm does not use SIMD instructions and does not
rely on the branch predictor to get a good performance, thus is good as
a general, default, architecture-agnostic implementation.
There is still a bypass for ASCII-only strings to benefits from
auto-vectorization, if the target supports.

Author: oxalica