Implemented a JSON5 lexer, as requested in #1880.

This lexer was mostly derived from the jsonnet lexer, which already implements lexing for features such as unquoted and single quoted object names and string values, comments and trailing commas in objects and arrays. This lexer covers the other JSON5 features such as mulitline strings, hexadecimal numbers, exponentials, leading and trailing decimal points, optional plus/minus signs and Infinity/NaN keywords while removing most of the jsonnet-specific code.

I'm not familiar with pygments or regex, so any feedback would be appreciated.