Bug fix on String type InnerVal encoding to use UTF-8 charsetName.

Current master's test failed on following environment and I think it is because of usage of getBytes without specifying encoding on InnerVal class.

More specifically, I think String's getBytes without encoding cause different result.

 case s: String =>
        val pbr = new SimplePositionedMutableByteRange(s.getBytes.length + 3)
        val len = OrderedBytes.encodeString(pbr, s, order)
        pbr.getBytes().take(len)

I am suggesting change String type InnerVal's serialize/deserialize to use UTF-8 encoding so running environment do not break test cases.