This has got me thinking (which never turns out well). So I've gone back to a random K100D PEF file from 2012 and compared how I processed it then with how I'd do a basic PEF to JPEG conversion nowadays.
Here's what I did with it back in 2012:
And here's what I did with it this morning using the standard approach I apply nowadays:
So yes, the differences I mentioned in my post higher up on this page are probably down to changes in my approach to processing rather than sensor differences. And when I look at these two samples, I have to conclude that in many ways my processing has become much, much WORSE over the past 7 years.
I think I need to reconsider what I'm doing with my raw files. In particular, to remember that I DON'T always have to recover as much shadow detail as possible, and that actually leaving quite deep shadows in a photo looks good.