Draft: Strings need converted to unicode when creating multipart request

[philtweir]

At two points in the multi-part request creation utility, encoding is missing to allow bytes-consuming functions to run.

utils.py:191

md5(datetime.now().isoformat()).hexdigest()

as DateTime.isoformat() will return a UTF-8 string, and

utils.py:224

body.append(b64encode(payload['data']))

as etree with return a UTF-8 string also.

The PR'd code is a draft to resolve this - if the general approach is considered appropriate, we can update tests/docs and refine for inclusion.