Stop adding whitespace in html that is not in the source document

[Zegnat]

This seems to be an under-documented feature of DOMDocument::saveHTML. It may sometimes add a \n at the end of its output. So when you just concatenate the string outputs of this method you may be introducing line breaks that weren’t in the original source.

I think adding a trim of some sort is wrong, as you might then also be trimming Text nodes that actually should contain the line break.

Instead what I have found to fix this is to move all the nodes into a DocumentFragment and retrieving the HTML of this fragment in one go.

---

Prior to this fix, the parser returns the following, note the \n in ["properties"]["content"][0]["html"]:

<div class="h-entry"><div class="e-content"><p>1</p><p>2</p></div></div>

{
  "type": [ "h-entry" ],
  "properties": {
    "content": [ {
      "html": "<p>1</p>\n<p>2</p>",
      "value": "1 2"
    } ]
  }
}