Cleanup interface and code, add new base16 decoder, etc

[k0ekk0ek]

This PR removes exposing the type and field information externally, thereby simplifying the code. It also adds the High performance base16 decoder from client9. Most importantly, it simplifies the SIMD and fallback scanners by adopting a slightly different interface and updates a lot of individual parser functions.

A lot of the code was more-or-less:

parse_xxx_rdata(...)
{
  // lex called in parse_rr
  if ((error = parse_yyy(...)) < 0)
    return error;
  lex(parser, token);
  if ((error = parse_zzz(...)) < 0)
    return error;
  ...
}

First the lex function would classify the type of token (e.g. <contiguous>, <quoted>, <line-feed> or <end-of-file>) to correctly set pointer and length, after which the parse_yyy and parse_zzz functions would check again to ensure proper syntax. In zone files though, often times we know exactly what type to expect and I've opted to remove the checks from the parse functions and simply do them in one go:

parse_xxx_rdata(...)
{
  if ((error = have_contiguous(...)) < 0)
    return error;
  if ((error = parse_yyy(...)) < 0)
    return error;
  if ((error = take_contiguous(...)) < 0)
    return error;
  if ((error = parse_zzz(...)) < 0)
    return error;
  ...
}

Leading to ~4% faster code.

Needs more testing etc before merging to main.