Well yes, NAT might pose some additional constraints, but my main line of argument is that even in an alternative timeline where we never had the IPv4 address shortage and therefore no pressure to develop NAT because every device can be assigned its own address just as is possible now with IPv6, we might still have ended up with default-deny-inbound firewalls for home networks anyway, because it might have turned out that letting random programs run world-accessible serves on random computers without any special user authorisation isn't such a good idea.
IPv6 doesn't require NAT, but my bog standard home router still firewalls it, and I need to manually allow inbound connections (or give up and just use UPnP).
Yes. Default-deny-inbound firewalls are much better than NAT because they are meant to provide security, can be managed (the "default"), and don't prevent deployment of new protocols. Also we might have a standard way to manage the firewall policy in this alternative future (upnp was done outside the IETF, and is a tire fire, because IETF rejected NAT).
But also the whole posture of a home network and consumer os might have been different without NAT, maybe the host based firewall would have won out, who knows. In the alternative universes we can't assume other things remain the same.
IPv6 doesn't require NAT, but my bog standard home router still firewalls it, and I need to manually allow inbound connections (or give up and just use UPnP).