The POST data contains the input twice, as testData, and testData2. However, when posting UTF-8 strings, the data is sent incorrectly.
1. Hit enter and note the second value misses two characters at the end.

Appending a dummy string (e.g. &dummy=12345678 to the xmlHttpRequest.send() variable is a workaround. Although data is still missing in the dummy variable.
2. Hit enter and see testData2 appear, and dummy now missing two characters.

This workaround is of course barely useful as the total number of bytes missing is equal to the total number of UTF-8 characters used, which could be all of them in some applications. Each UTF-8 character added to the input removes two more characters of dummy (and eventually testData2). It seems like at some point, xmlHttpRequest is igoring all but one byte of UTF-8 characters (in this example triggered twice).