/*
  The first 256 entries are tuples of 4 bit values. The lower bits
  are a mask that when xor'd with a byte removes the leading utf-8
  bits. The upper bits are a character class number. The remaining
  160 entries are a minimal deterministic finite automaton. It has
  10 states and each state has 13 character class transitions, and 
  3 unused transitions for padding reasons. When the automaton en-
  ters state zero, it has found a complete valid utf-8 code point;
  if it enters state one then the input sequence is not utf-8. The
  start state is state nine. Note the mixture of octal and decimal
  for stylistic reasons. The first 128 entries are obviously some-
  what unnecessary, but removing them would require another branch.
*/