During parsing, we sometimes need to get the string on base 10 of a BigInt, but Lexer only provides us `bigIntString` and `radix` as information. It means that whenever radix is not 10, we need to do some computation to get the String that represents a BigInt on decimal base. Current approach is using JSBigInt to perform this operation, but this is a very naive and unoptimized. We should investigate better options there.