Welcome to the Q&A Forum

I can't replicate this in live SERPs, but Matthew's explanation seems to match the evidence. It does appear that Google is doing some kind of implicit canonicalization. It's rare, but it does happen. You could try giving the PDF a unique title in its own meta information (the document doesn't seem to have any). It's more likely to look like a duplicate if that's undefined. The next step would be to recreate the PDF slightly and make it unique, but honestly, each step is going to get to be progressively more work. It really depends whether this is just an anomaly or it's causing you actual issues as a business.

You can try the service of Salesforce with the help of Axis consulting. I have tested it on my site.