Avoid cutting off utf8 encoding halfway when truncating text values

[karimbahgat]

Currently encodes any unicode to bytes (using whichever encoding), and truncates afterwards to fit within the specified byte size of the field. I guess UTF8 and any encoding that allows characters that spans more than a single byte would stand at risk of cutting and invalidating such characters if they occur at the end of the text, spanning across the truncating limit. See #125.