Fix premature SNBT parsing as number

[kennytv]

┬┴┬┴┤(･_├┬┴┬┴

If a numeric token is found before an unquoted string's end, it will be prematurely parsed as a number and errors because of remaining/unread characters for strings such as 0dd, 11fa, or 1350diamonds_plz.

Instead of constantly looking for the numeric tokens, this just reads the full string and if all chars meet the required criterea (previously possiblyNumeric) except for the very last character, the token parser will be called. If it's possibly numeric all the way through, it goes to the int parser, otherwise to the same code below.