Why they tend not to do it is is likely because of infrastructure, privacy and security concerns, regulations (EU privacy laws come to mind). And furthermore, they'd need to find a good home for it, or risk further backlash. So there are risks in terms of legal, regulation, and resources. A company like Google can't just transfer SSH keys to the highest bidder on eBay.
All that said, I think in this case, there might be a more compelling reason to make the effort than normal. There's clearly a lot of passionate users, many of them influencers, and it's relatively simple to transfer people's OPML with some extra history data. I can see it's far from trivial though.
Could they, even? They don't build isolated products the same way that many other software companies do. Everything they build is built on their own infrastructure, not some commodity stack. Even if they released the source code to Reader for free, would it do anyone any good without the full suite of BigTable/Chubby/GFS/Googlebot/etc. services that I'm sure it is tightly integrated into?
"If the company maintained every niche product with N thousand fans, even paying ones, it’d become the very bungling bureaucracy we love to hate. For a company with Google’s ethos and standing, any such dead-end, non-revenue-producing product that’s retained is holding others back"
For any future CEO's of large companies that might be reading the above please keep in mind that sop for this situation is to sell the product or service to another company. This has typically been how business has always operated. Even if the business lost money there is someone out there who would buy it. Taking the "don't have time for this we are moving so fast" is nonsense. The "distraction" and "rounding error" card is way overplayed.