UnitSentry.file_contents() breaks for non-utf8 files

[Björn Tillenius]

If you try to get a binary file using UnitSentry.file_contents, if will break with a
UnicodeDecodeError, since it always tried to decode the contents using UTF-8.

Traceback (most recent call last):
  File "/usr/lib/python3.4/unittest/case.py", line 57, in testPartExecutor
    yield
  File "/usr/lib/python3.4/unittest/case.py", line 574, in run
    testMethod()
  File "/home/bjorn/projects/landscape-charm/trunk/tests/basic/test_service.py", line 126, in test_error_pages
    haproxy_path + http_code + ".http", "haproxy/0"),
  File "/home/bjorn/projects/landscape-charm/trunk/tests/helpers.py", line 94, in get_file
    return unit_sentry.file_contents(path)
  File "/usr/lib/python3/dist-packages/amulet/sentry.py", line 81, in file_contents
    output, return_code = self._run('cat {}'.format(filename))
  File "/usr/lib/python3/dist-packages/amulet/sentry.py", line 118, in _run
    return output.decode('utf8'), p.returncode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x8b in position 72: invalid start byte

There should be a way to get binary files as well as text files.