I'm 99% sure it won't have much of a positive impact in sales simply because the majority of ebook buyers don't care about DRM anyway, only the minority of us bother to strip DRM. So while it wouldn't be a large drop in sales I do think it would still be a drop. It might not be enough of a loss for them to care, and tbh as you said it would probably only result in a mild increase in piracy while the majority either do library loans or switch to paper.

I don't think that those of us who care enough to jump through the hoops to strip DRM are just gonna roll-over and accept that the publishers can yoink our entire libraries whenever they see fit, but I do admit most don't care. However those that don't care aren't stripping DRM anyway, they are just relying on the ability to redownload their books whenever they wish from Amazon/Kobo/Nook.

with books there's basically no reasonable way to create an ebook from a hardcopy

On the contrary, tons of books have been digitized from hard copies through a combination of OCR and manual editing. (E.g.: Project Gutenberg.) The same basic process works for both printed books and pages displayed on an e-reader. It's quite tedious but not exactly difficult. Anyone with a smartphone can submit usable scans, though some simple DIY equipment speeds up the process and improves the quality, and OCR is getting better all the time.

In the worst case the book can simply be retyped. People used to copy books by hand after all, using nothing more sophisticated than pen/quill and paper/parchment/papyrus. Unlike in those days the manual effort is only needed once per title, not per copy.