v0.2.2

• Improve handling of texts with misaligned tokens.
• Add the ability to use token.feats as tags.
• Get language attribute directly from Stanza pipeline object.

Thanks to @yuhaozhang and @SergeyShk for the pull requests.