PDF export v2 ignores subsequent query strings of images added via URL

Issue Summary

• When we have images with the same base URL, but with different query parameters added into a Confluence page, the v2 PDF export caches the image from the first query parameter, and returns the same image in the export.

Steps to Reproduce

1. Create a new page in Confluence, in either the legacy, or current editor.
   #Use the storage format below for the page;

   • <ac:image>
     	<ri:url ri:value="https://dummyimage.com/600x400/000/fff.png?text=a" />
     </ac:image>
     <ac:image>
     	<ri:url ri:value="https://dummyimage.com/600x400/000/fff.png?text=b" />
     </ac:image>
     

   • If you're on the legacy editor, you can also add the image URLs directly using the "Insert from the web" option when adding attachments.
2. The above should display two separate images;
   • One image with the letter "a"
   • And another with the letter "b".
3. Export the page as a PDF

Expected Results

• We should see two distinct images in the PDF export, similar to what we see on the page.

Actual Results

• The first image is used for the second image as well.
• It appears that the image is queried once, and re-used for the subsequent images.
• This issue is not found with the older PDF export option.

Workaround

• Currently there is no known workaround for this behavior with the new PDF export service.
  • A workaround will be added here when available
• Alternatively, we can force the v1 PDF service to run;
  • Either by enabling IP Allowlisting (site wide), or
  • Adding a blank stylesheet tag for the PDF export in the Space settings (space wide)