Reduce lexer run-time costs. See also work done in #338 for the *parser*; now that work should continue into the *lexer* to achieve maximum 'parse' speeds.