Hello,
with home connections moving beyond gigabit the time of 30€ routers plastic routers seems to end slowly but surely.
Some routers have eMMC, some microsd cards and some even m.2 slow from which they can boot. So I would propose to make the update mechanism of OpenWRT more solid. Attended Sysupgrade were a good step in the right direction but it is not enough for the normal user which want to setup ones and then forget and also for the advanced user which like to tinker with things but does tinker not only with his OpenWRT device but also others.
I would see 2 things that would help a lot in that regard. Switch to BTRFS with schnapps shipped for the device on target where lowend hardware simply does not exist, like eMMC by default and a gig of RAM or whatever you guys would see as baseline from which on doing that change would make sense.
Also I would see a need for unattended upgrades which could be implemented safely with the switch above. Well actually the need for unattended upgrades is there for decades by now but there was no good way to implement it I guess because of the limits of squashfs and the wide hardware support.
I would like to hear feedback if you would see a benefit in the changes and what would be challenges for it.
Realistically, that's not going to happen. It just isn't as easy to implement, as you seem to think - and anything (substantial) done on this, would directly hurt the primary targets which don't have much storage to spare and usually very limited bootloader features.
EDIT: btw., there is quite a lot of potential to improve x86_64 support incrementally (evolutionary, rather than revolutionary), especially in terms of bootloader, dual-boot, partitioning and maybe even snapshot support (as well as reducing the 32 bit x86 targets to one) - I'm not so sure working on btrfs specific features (and thereby depending on a single filesystem, which is quite different from more 'normal' linux filesystems) would be the smartest option though. So if you have time to kill, there are quite a few opportunities.
The change does not apply to low end targets, it is more for the long run especially to have vendors leaverage that features. There are to many that take a openwrt base then develope their hardware arround the need of that base. OpenWRTs small storage footprint is actually one of the reasons why there is no need for vendors to put more and robust storage on the device. The time where price would have prevented that are over because I really doubt that the 0,30$ on a 300-500$ router are the driving factor not to do it.