HTML reader: parse footnotes using dpub-aria roles

[jgm]

Using the dpub-aria roles doc-endnotes, doc-endnote, doc-noteref, and doc-backlink, we should be able to parse footnotes from an HTML document into pandoc Notes.

See #4213.

[Seirdy]

Agreed that this would be a great feature to have. Note that doc-endnote is deprecated as of DPUB-ARIA 1.1.

[jgm]

A note reference looks like

<a href="#fn1" class="footnote-ref" id="fnref1" role="doc-noteref"><sup>1</sup></a>

The notes:

<section id="footnotes" class="footnotes footnotes-end-of-document"
role="doc-endnotes">
<hr />
<ol>
<li id="fn1"><p>there<a href="#fnref1" class="footnote-back"
role="doc-backlink">↩</a></p></li>
</ol>
</section>

When pandoc parses the note reference, it already includes the role:

    [ Link
        ( "fnref1"
        , [ "footnote-ref" ]
        , [ ( "role" , "doc-noteref" ) ]
        )
        [ Superscript [ Str "1" ] ]
        ( "#fn1" , "" )
    ]

The information is also present in the parsed footnotes section:

, Div
    ( "footnotes"
    , [ "section" , "footnotes" , "footnotes-end-of-document" ]
    , [ ( "role" , "doc-endnotes" ) ]
    )
    [ HorizontalRule
    , OrderedList
        ( 1 , DefaultStyle , DefaultDelim )
        [ [ Div
              ( "fn1" , [] , [] )
              [ Para
                  [ Str "there"
                  , Link
                      ( ""
                      , [ "footnote-back" ]
                      , [ ( "role" , "doc-backlink" ) ]
                      )
                      [ Str "\8617\65038" ]
                      ( "#fnref1" , "" )
                  ]
              ]
          ]
        ]
    ]

So it shouldn't be too hard to define an AST transformation that runs after the main HTML parsing and converts the note-links to proper Note elements. There is no need to change the actual parsing.