Curses-independent but complete mulitbyte sequence input handling

[jquast]

Per the UP/DOWN/LEFT/RIGHT detection procedure part of,

https://github.com/ianepperson/telnetsrvlib/blob/master/telnetsrv/telnetsrvlib.py#L623

I have implemented full multibyte sequence detection for terminals of all types, given a 'term' string. It imports curses, but it never calls the nasty init screen routines that garbage up the terminal. It is available as a decoder function to the fork of the 'blessings' module in my github account. These changes will be incorperated into the next version of blessings. You can see a demonstration here:

http://ascii.io/a/2859