I can understand the reasoning behind the move, given the goal of attempting to keep NetBSD relevant as a research and embedded systems OS. While NetBSD certainly provides more value in the "cross-platform hobby OS" niche than anything else out there it's not a niche that pays the bills with research grants, etc. There may well be something the core developers have in mind for future of the network stack or storage architecture that they're concerned just won't naturally scale downward enough to fit on some of the more resource-constrained systems on the supported list. (And there's also starting to be toolchain/compiler support issues with some of the more obscure architectures.)
It may be perfectly fair for the core team at this point to draw a line in the sand and say "Look, we need to stop wasting the core developer's time on fixing collateral damage that only affects 'pet distributions' with tiny user bases. We'll support the users of those distributions however we can in helping *them* develop fixes and commit them, but we can't hold the evolution of the mainline OS hostage to ports that can't summon even a single committed developer." I do wonder, though, if that's really going to be enough to make NetBSD a hotbed of innovation again. I haven't followed the BSD scene that closely for several years, but it really seems to me that FreeBSD has stolen a lot of bleeding-edge thunder from NetBSD. (While OpenBSD keeps on doing what it does best, consuming its own arse-end like a never-ending Ouroboros serpent while still managing to produce some decent universally-applicable security code.)