etcd 3.5 members manipulation & change of healthcheck

[camaeel]

When I was using etcd version 3.4 it was possible to manipulate etcd members (for example list) when the cluster lost forum and fix some situations when it was broken. After I upgraded to v3.5.12 which contains this change I can't execute etcdctl members list on a node that lost quorum. Is there maybe some workaround for that?
If not, then maybe a etcdctl flag to use local members configuration could be added, so in case of disaster maintainer could be able to fix members list manually?

Also now /health endpoint is responding with "ok" only on a cluster that has quorum. Is there maybe an option to still use /health with old behaviour (maybe with some http request parameter)?
I know there are new endpoints /livez and /readyz provided in 3.5.12, and /livez seems to behave like old /health did.

[camaeel]

Update:

I found workaround for getting member list I can use curl command: curl -XPOST localhost:2379/v3/cluster/member/list | jq

But it is not possible to execute member remove this way: curl -XPOST localhost:2379/v3/cluster/member/remove -d '{"ID": "14663967015014024213"}'. I get a timeout.