Unicode event header/body lengths are not correctly calculated

[sgnn7]

Currently when sending an event using unicode characters, the length of the string sent to statsd is evaluated using codepoints, causing string clipping in the final data ingested. Also, Java uses UTF-16 and the agent expects UTF-8 so we have to use the UTF-8 calculations explicitly.

AC:

• UTF-8 char[] length is used as the sent length of statsd event headers/message.