Fix for issue #50, to ignore errors on shiftjis encoding

[rheenen]

…when trying to detect the type of contents.

[heuer]

Sorry, wasn't aware that comments to a forked repository contribute to a PR as well. It wasn't my intention to hijack this PR

[heuer]

Suggestion (got errors with Py 3.4 / 3.6 within another project):

def encode(self, encoding=None, errors='strict'):
    if encoding == 'shiftjis':
        raise LookupError("unknown encoding: shiftjis")
    return super(FakeString, self).encode(encoding, errors)