Allow fenced divs and bracketed spans to have attributes with no value

[jimjam-slam]

Currently, fenced divs and bracketed spans in Markdown can have general attributes (not a class or ID) that get prefixed with data- in the resulting HTML output. However, these attributes must have a value (even if that value is just an empty string), or else Pandoc skips div/span processing altogether:

---
title: "Attribute test"
---

Here is a span with a value:

[Hello]{my-greeting="hello"}

And here is one without a value. It doesn't seem to be recognised as a bracketed span at all:

[No value]{my-greeting}

This one has two attributes. `my-greeting` has a value but `my-farewell` does not:

[Hello but no goodbye]{my-greeting="hello" my-farewell}

Now a span with attributes, both of which have a value:

[Hello but no goodbye]{my-greeting="hello" my-farewell="bye"}

And finally, a span where `my-farewell` has a value that is an empty string:

[Hello with empty goodbye]{my-greeting="hello" my-farewell=""}

Is this true for fenced divs too?

:::{.myclass my-greeting="hello" my-farewell}
Some content wrapped in a div
:::

On Pandoc 3.2:

pandoc -f markdown+bracketed_spans -t html -o index.html index.md

<p>Here is a span with a value:</p>
<p><span data-my-greeting="hello">Hello</span></p>
<p>And here is one without a value. It doesn’t seem to be recognised as
a bracketed span at all:</p>
<p>[No value]{my-greeting}</p>
<p>This one has two attributes. <code>my-greeting</code> has a value but
<code>my-farewell</code> does not:</p>
<p>[Hello but no goodbye]{my-greeting=“hello” my-farewell}</p>
<p>Now a span with attributes, both of which have a value:</p>
<p><span data-my-greeting="hello" data-my-farewell="bye">Hello but no
goodbye</span></p>
<p>And finally, a span where <code>my-farewell</code> has a value that
is an empty string:</p>
<p><span data-my-greeting="hello" data-my-farewell="">Hello with empty
goodbye</span></p>
<p>Is this true for fenced divs too?</p>
<p>:::{.myclass my-greeting=“hello” my-farewell} Some content wrapped in
a div :::</p>

The HTML spec allows absent values for attributes — in fact, boolean attributes (such as loop on videos) are always true if present (ie. myattribute and myattribute="false" are both true, not false; for the latter you must omit myattribute altogether).

In this case I'm talking about the custom attributes that get prefixed with data- on the way through Pandoc, not the specified boolean attributes in the HTML spec, but I think there are plenty of use cases where someone might want a custom attribute without a video. Lining up with the spec also helps users with HTML experience inuit what they may be able to do.

For example, in qmd-lab/closeread#139 @andrewpbray and I are allowing users of our Quarto extension to play and pause videos with spans, and we're deciding between [my trigger](.play-video), [my trigger](play-video) and [my trigger](play-video="true") as user APIs. The second is most intuitive to my mind (although I accept that may be down to taste).

(Potentially related issue: #8251 addresses the class case specifically, where it's desirable to remove that attribute when it has an empty value.)

[jimjam-slam]

Doing some more reading and realising that attributes are always supposed to have a value if they are not one of the specific boolean attributes used in HTML. So maybe this is a bad idea after all!