[upstream] implement extracts using Parsoid

[waldyrious]

The extracts are currently generated with the old API which doesn't produce valid xhtml5, so any malformed ajax request just won't show up in the content area.

Parsoid does produce xhtml5, so Extension:TextExtracts needs to be extended to support parsing with Parsoid rather than the old API. Trail of bug tracker links (updated as of 2017-11-20):

• ~https://bugzilla.wikimedia.org/65169~ original request: "TextExtracts: Support for Parsoid-based extracts"
• ~https://phabricator.wikimedia.org/T67169~ same issue, migrated to Phabricator
• ~https://phabricator.wikimedia.org/T113094~ "[EPIC] The Page Summary API needs to provide useful content for the majority of articles"
• ~https://phabricator.wikimedia.org/T166272~ sub-issue of the above. "HTML version of text extracts is not balanced/well formed and naive". Closed, declined:

  I have made the decision that this will not be fixed. T170617 will make sure that api consumers know about this problem. For those who want well formed HTML we will be providing a new service on RESTBase which will guarantee that (please follow along with T113094)

• https://phabricator.wikimedia.org/T165017 AC (acceptance criteria?): "[...] The extract property is present and contains valid HTML." Closed, resolved. Does this mean the functionality is now available?
• https://phabricator.wikimedia.org/T177431 it looks like this is the task whose completion will enable well-formed HTML extracts (extract_html in the example output). Note: isn't that already available?

[waldyrious]

Update: from Tech News issue #11, 2018 (March 12, 2018):

Page Previews has been updated to use HTML for previews. (...) [2]

[waldyrious]

See also the code of Module:Excerpt and Navigation popups (source code).