The Import-book-as-DOCX method should not be doing anything different than the DOCX Conversion, except for preserving styles and stuff.
I assume the Chapters are properly referenced within the book, and that a conversion generates the appropriate pagebreaks...
Here's a thought.
The conversion pagebreaks are probably created automatically, using the Structure Detection part of the Conversion process. I don't know if the DOCX Import does Structure Detection.
Now, if the pagebreaks were defined in the DOCX itself, rather than created by calibre conversion (which assumes any Header style with text that contains "Chapter" is... a chapter, and makes a pagebreak) -- then DOCX Import should handle that properly.
Fortunately, you can still use Structure Detection from the editor. Structure Detection uses XPath, and you can probably just copy the value in the Conversion dialog and use it in the editor.
Just right-click in a file and select "Split at multiple locations".