Review unicode support

Unicode support was added somewhat hastily. I pretty much just shoved a b in front of all the strings, and switched StringIO for BytesIO.

I'm not entirely sure that StringIO is even used anymore. The import may be removable
For some reason, StringIO can be imported as either from StringIO import StringIO or from io import StringIO, what are their differences? does something similar apply to BytesIO?

This code

nrepl-python-client/nrepl/bencode.py

Lines 106 to 109 in a97ba25

    
           # x = x.encode("UTF-8") 
        
           # TODO revisit encodings, this is surely not right. Python 
        
           # (2.x, anyway) conflates bytes and strings, but 3.x does not... 
        
           out.write(str(len(x.encode('utf-8'))).encode('utf-8'))

the comment probably doesn't need to be there anymore and it should be split up to ensure that it's understood that it's doing two things:

Getting the length of the string in bytes
Turning that length into a utf-8 encoded string

There's probably more refactoring to look into, but this is what comes to mind.

	# x = x.encode("UTF-8")
	# TODO revisit encodings, this is surely not right. Python
	# (2.x, anyway) conflates bytes and strings, but 3.x does not...
	out.write(str(len(x.encode('utf-8'))).encode('utf-8'))