Good evening,
after some testing it seems that I am just a very stupid person.
The reason, why the interface is always down is - after I had a look into the source code of netcfg - that is simply fails *and*, before it fails, netcfg would always reset the link, remove any routes and ip addresses. At least that is what I assume.
But why can I manually configure the network? Because I used the onlink flag:
# ip link set ens1 up
# ip addr add 255.255.8.243/29 dev ens1
# ip route add default via 255.255.8.1 dev ens1 >>>onlink<<<
"onlink: pretend that the nexthop is directly attached to this link, even if it does not match any interface prefix."
https://linux.die.net/man/8/ip
The /29 subnet actually does not exist on the host; in reality, it is a /24 subnet. While /etc/network/interfaces in the installed Debian system itself would be aware of whatever it is aware of - I just don't know - and automatically will configure the route using the onlink tag, in the debian installer that is not possible. It seems the only correct way would have been to configure the network with the /24 notation and the according 255.255.255.0 netmask, which makes sense, because working with a /29 notation using MacVTap and VEPA does not make much sense for me if the subnet is configured as /24 on the host.
However, I am not well-educated when it comes to networking at all. I think my assumptions are right, but if the debian installer should be aware of that (like the final Debian installation is), we should leave that issue open.
--etkaar
