Option to avoid deleting the kernel_ll address from bridges.

[robmry]

- What I did

As discussed in this Slack thread, replacing the kernel-assigned link local address on bridges may be causing issues.

It'd be better not to replace it, and there shouldn't be any consequences but, to avoid the risk of a breaking change in a patch release - this PR introduces an environment variable that only modifies the current behaviour when set.

The kernel-ll address has been removed since #46850 - part of an attempt to prevent daemon startup failures following changes in fixed-cidr-v6. But, by treating the standard LL prefix as a special case, removal of the kernel-assigned LL address can be avoided.

This will make it possible to experiment and hopefully work around the problem. In 27.0 we should remove the env var and make the new behaviour the default, or revert this change - #47778.

- How I did it

If env var DOCKER_BRIDGE_PRESERVE_KERNEL_LL=1, don't assign fe80::1/64 to a bridge, and don't delete any link local address with prefix fe80::/64.

- How to verify it

Modified regression test, just to make sure setting the new env var doesn't cause any startup problems for the default bridge, and that it still ends up with a link-local address.

Without the env var set, for a user-defined network ...

docker network create n6 --ipv6 --subnet fdaa::/64
docker run --rm --name c1 -dti --network n6 alpine

12: br-6ae7b0edf268: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 02:42:07:85:54:ba brd ff:ff:ff:ff:ff:ff
    inet 172.19.0.1/16 brd 172.19.255.255 scope global br-6ae7b0edf268
       valid_lft forever preferred_lft forever
    inet6 fdaa::1/64 scope global tentative
       valid_lft forever preferred_lft forever
    inet6 fe80::42:7ff:fe85:54ba/64 scope link tentative
       valid_lft forever preferred_lft forever
    inet6 fe80::1/64 scope link tentative
       valid_lft forever preferred_lft forever

dockerd restart ...

12: br-6ae7b0edf268: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default
    link/ether 02:42:07:85:54:ba brd ff:ff:ff:ff:ff:ff
    inet 172.19.0.1/16 brd 172.19.255.255 scope global br-6ae7b0edf268
       valid_lft forever preferred_lft forever
    inet6 fdaa::1/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::1/64 scope link
       valid_lft forever preferred_lft forever

With the env var set ...

Initially, and after daemon restart ...

17: br-e476bcfaa2de: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 02:42:c8:2b:c2:4a brd ff:ff:ff:ff:ff:ff
    inet 172.19.0.1/16 brd 172.19.255.255 scope global br-e476bcfaa2de
       valid_lft forever preferred_lft forever
    inet6 fdaa::1/64 scope global tentative
       valid_lft forever preferred_lft forever
    inet6 fe80::42:c8ff:fe2b:c24a/64 scope link tentative
       valid_lft forever preferred_lft forever

- Description for the changelog

Experimental environment variable `DOCKER_BRIDGE_PRESERVE_KERNEL_LL=1` will prevent the daemon from removing the kernel-assigned link local address on a Linux bridge.

[vvoland]

Behavior stays the same without an explicit opt-in via the env var, so LGTM