support open encoding utf-8

[nooperpudd]

No description provided.

[Byron]

Thanks for your contribution. I assume this makes all refs work, not just those that are ascii compatible.

[brunson]

Seems like it could use some unit tests.

[Byron]

Definitely, and they should be easy to manufacture too. I hope it will be contributed eventually.

In this case, specifying utf-8 directly should be backwards compatible, while allowing for utf-8 encoded bytes to be interpreted correctly. The default encoding is ascii in default python installations, which is naturally included in utf-8 encoding.