I have an implementation of jetscii searching for aarch64 neon simd that I'm working on, which is looking to be very competitive (it's currently looking nearly always faster than memchr3, but supporting more than 3 bytes to search for 👀). See also #60, which is necessary for full speed iterating over the indexes of the bytes we're looking for.

@shepmaster I know this repo isn't super busy these days, would you be able to accept such a PR?