The workflow: You develop your image from raw into large gamut color space (even larger than AdobeRGB) and this is your source. The photo editor then handles automatically and behind the scenes conversions for your display or your printer. Or you can export sRGB for web.
I just read an article by a photographer who explained that he always works in AdobeRGB because it has a larger gamut. This is fair enough if his monitor can display it...
It does not matter what gamut the monitor has. That's different data path and you don't even do anything about it. It is handled by the color management system for you automatically and only that data floating to your monitor are reduced in gamut. Nothing else. Your original stays in large gamut. Even if your monitor is sRGB, you can print larger than AdobeRGB, because you print from the original data that is large gamut.
...but surely it isn't possible to get a 'better' print by starting out with a larger gamut and squeezing that more to fit the paper range? If parts of sRGB images are out of gamut for print, why would it be preferable to print from AdobeRGB?
The gamut shape of printer/paper/ink combination has different shape than the synthetic sRGB or AdobeRGB profiles. In some areas it is smaller, in some areas it is larger. My oldish Epson 3800 can print colors that are outside AdobeRGB gamut. If I reduced the image to sRGB before printing, I would lose some nice blues and greens. There are areas that the printer can't print, but this is routinely solved by profile conversion or if the automated process is not good enough for you, you can use soft proofing and correct individual out of gamut colors manually.