pub fn tokenize_words(text: &str) -> Vec<&str>
Simple word tokenizer that splits on non-alphanumeric characters.
Produces only non-empty tokens of alphanumeric characters.