Fix byte-encoding of Strings / Unicode safety

[msakrejda]

Currently, the {read,write}UTF{,Bytes} methods are used. This may not always be sane.

[msakrejda]

For instance, I think length calculations in message encoding are almost certainly bogus, since they're occuring in terms of the number of characters, not the number of bytes.

A possible workaround here is to write the string out to a byte array and see how many bytes we wrote, but this seems rather silly (and hopelessly inefficient). Nevertheless, there may not be a better alternative at this time.