On finding characters within a string in an optimized way.

Due to the fact how current CPUs work, sometimes the best optimization appears counterintuitive. I still haven't internalized the fact that either manual vectorization or using SIMD intrinsics is better for performance than naive loops, especially on languages that have some interface with low-level stuff. (Unfortunately, more often than not I get to spend the time thinking about this while working within PHP which has plenty of barriers preventing such optimizations from being of any use.)