Mixed-version robustness test failing

This is an AI assisted analysis, robustness tests look to have been broken https://prow.k8s.io/job-history/gs/kubernetes-ci-logs/logs/ci-etcd-robustness-main-amd64 since the raft bump and https://github.com/etcd-io/etcd/pull/21846 and https://github.com/etcd-io/etcd/pull/21857 has not fully fixed them.

### Analysis

`ci-etcd-robustness-main-amd64` fails in mixed-version (3.6 + 3.7) runs when merging per-member WAL history. The bootstrap ConfChange entry at raft index 1 is reported as mismatching even though the members agree on the conf change. The only differing bytes are `\b\x00` (`ConfChange.Id` encoded as 0), which a 3.6 member writes explicitly and a 3.7 member omits.

Run: https://prow.k8s.io/view/gs/kubernetes-ci-logs/logs/ci-etcd-robustness-main-amd64/2062255973065560064

```
EtcdHighTraffic/ClusterOfSize3/MemberReplace (etcd 3.6.12 + 3.7.0-beta.0):
main_test.go:113: mismatching entries on raft index 1, diff:
"Data": bytes.Join({
- "\b\x00",
"\x10\x00\x18...{"id":12416079282240904009,...}
}, ""),
"Type": s"EntryConfChange",
```

Note: prow shows this as flaky because `--count 120` reruns the same scenario names and only some iterations hit the mixed-version MemberReplace path. The failure is deterministic on that path.

Start at main_test.go:113 and reproduce the mixed-version MemberReplace failure with the linked ci-etcd-robustness-main-amd64 run. Compare the raft index 1 ConfChange entries written by etcd 3.6 and 3.7, then verify the mixed-version robustness scenario completes without a mismatch.

Written by the indexing model from the issue text.