Set IsReplicating on APPENDLOG init handshake to prevent idle resync loop

[nattress]

After a TCP disruption on an idle cluster, the replica's EnsureReplication
would trigger repeated resyncs because IsReplicating was only set when a
data APPENDLOG arrived, not on the init handshake (-1,-1,-1). On an idle
stream no data APPENDLOG follows, so the flag stayed false and
EnsureReplication fired every ClusterReplicationReestablishmentTimeout.

Move IsReplicating = true to the top of NetworkClusterAppendLog so it is
set on every APPENDLOG message including the init handshake. The flag is
cleared implicitly when the session is disposed on connection drop.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adjusts replication-session semantics so IsReplicating is set for the replication stream starting with the APPENDLOG init handshake, preventing unnecessary resyncs when the stream is idle.

Changes:

• Updated IClusterSession.IsReplicating XML doc to reflect “active replication stream” semantics (incl. handshake).
• Moved APPENDLOG “initialization message” handling into the replica-path and sets IsReplicating = true before the init handshake returns OK.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
libs/server/Cluster/IClusterSession.cs	Clarifies IsReplicating contract to include init handshake as part of active replication stream.
libs/cluster/Session/RespClusterReplicationCommands.cs	Marks sessions as replicating earlier and handles init handshake in the replica branch to avoid spurious resyncs.