After fixing #13286 we'll end up with internal image/link templates producing invalid markup for .PlainText with HTML entities, e.g:

![A's is > B's](sunsets.jpg)

Will become:

<img src="sunsets.jpg" alt="A&amp;rsquo;s is &amp;gt; B&amp;rsquo;s"></p>

That renders to

Assuming the image does not exist.

So, to fix this, we can either:

1. Pipe PlainText to safeHTML in our internal templates.
2. Make PlainText() return template.HTML.

What do you think, @jmooring ? I'm fine with both options. Just to be clear: Goldmark produces the exact same markup for the example above (when markup.goldmark.renderHooks.image.disableDefault=true), but since Content returns template.HTML, that's not an issue.

You could argue that the above is a bug in Go, and I would agree. It should consider HTML entities in its escaping.