This release brings a ~24% performance improvement in existing benchmarks for hck's single-byte-delimiter same-field-order parser (aka fastmode).
Essentially, hck was using memchr2 to look for newlines and separators at the same time and would continue to use memchr2 to parse a line even after the right most field was found. Now, after the rightmost field is found it switches to just memchr and searches for the next newline, effectively short-circuiting the line.
On workloads that just use the first couple fields of a file this will be a massive performance improvement.