They measured a speedup. SIMD is a super common way to speed stuff up and good compilers do it automatically.
A lot of string comparisons use SIMD now. In C++ if you use std:string on GCCit has done it for years. I bet Rust and Python (in the compiled runtime) ate doing it too.
No sense leaving easy performance gains on the table.
Keep on thinking. You never know what else you will come up with.