it wrongly matched `<w:pPr` when looking for `<w:p`
in that case the cloned block would not be valid XML,
and so the generated docx document could not be opened by Word
adding the `\b` word boundary to the regex fixes this.
the accompanying unit test failed with this message when trying to load
the generated file:
```
ErrorException: DOMDocument::loadXML(): Opening and ending tag mismatch:
p line 2 and body in Entity, line: 2
```
This might solve issues such as #681 and #664.