make combine_files() work with UTF-8 encoded files

[evgeni]

setup.py might fail on Python 3 in non-UTF8 environments while parsing
HISTORY.rst or README.rst which are not required to be ASCII-only with:

UnicodeDecodeError: 'ascii' codec can't decode …

We could also require the files to be plain ASCII, but it sounds not the
right thing to do in the 21st century.

Simple reproducer:

% LC_ALL=C python3 setup.py
Traceback (most recent call last):
  File "setup.py", line 126, in <module>
    long_description=combine_files('README.rst', 'HISTORY.rst'),
  File "setup.py", line 112, in combine_files
    file_contents.append(f.read())
  File "/usr/lib/python3.5/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 2132: ordinal not in range(128)

[coveralls]

Coverage remained the same at 89.986% when pulling 7566412 on evgeni:non-unicode-setup into 02c2b97 on ansible:master.

[AlanCoding]

This makes sense to me. I believe the idea is that files that we are providing are implicitly utf-8, and it just needs to be marked as such.

file -I HISTORY.rst 
HISTORY.rst: text/plain; charset=utf-8