string encoding/decoding is imcompatible with dubbo-hessian-lite for four char > 0xFFFF

[wbpcode]

According to the Hessian2 standard, the string should be UTF8 encoded. However, UTF16 string is used by the Java. And when the dubbo-hessian-lite encoding the string, every UTF16 char is treated as one UTF8 char which is wrong.

Considering that most of dubbo users using dubbo-hessian-lite, we should try to provide the similar support for compatibility, even it's wrong.

[wbpcode]

This needn't be part of the core lib. We can extend the new type support by the toHessian2/fromHessian2.

[wbpcode]

I will keep this open because others may encounter similar problem.

[wbpcode]

Seems only 4bytes utf-8 cannot be handled correctly, may be we can still fix it. 🤔