Presumably though that involves image layers and an image-layer-aware filesystem so that they don’t actually have to have more than one copy of the common image

Presumably though that involves image layers and an image-layer-aware filesystem so that they don’t actually have to have more than one copy of the common image per execution machine, and just have to send around the customer’s overlay diff? Maybe they don’t bother and just pass the whole thing around, but this is a solved problem in the wider world for distributing a large number of large OCI images that are substantially similar.