asccidoc_reader: Do not encode into latin1 on python3

[gileri]

It causes UnicodeDecodeError when using non-ascii characters in titles such as 'à'. The commented example (ñ) doesn't work in python3.7 with the encode/decode hack, but works fine without. I don't really understand why it was introduced.

I haven't run the test suite, I can't manage to get a working testing environment with pelican and pelican-plugins.