Fix rust binding to use the chars count instead of byte length

[arai-a]

The rust binding uses self.sym.chars().nth(self.index) to access chars, but the sym.index uses the [0, self.sym.len()) range.

chars().nth() uses the char index, while sym.len() is the byte length.
This results in InvalidIndex error in SymbolParser::current if the symbol contains non-ASCII range.

This patch introduces a new field sym_len to SymbolParser, which caches the result of chars count (given it's O(N) operation),
and use it in place of sym.len().

[jupblb]

Sorry for a late response!

[jupblb]

Thanks :)