Slicker does not correctly move definitions containing unicode

[benjaminjkraft]

If you try to move a single symbol whose definition contains unicode (and uses an imported name) to a file which does not already have a PEP 263 encoding comment, slicker crashes.

Here's a repro:

# test.py
# coding: utf-8
import x

def foo():
    x.y()
    return 'bár'

Then run slicker test.foo test2.

A workaround is to add the coding comment to the destination file in advance.