[BUG] [Segment Replication] Incorrect use of default codec and replication checkpoint

[dreamer-89]

Describe the bug
While working on #3881, identified couple of issues

1. During primary shard recovery, the force segment sync call use default codec implementation vs codec defined in engine config. This is problematic when non-default Lucene codec is used and will fail codec specific validations on primary
2. Target sends back ReplicationCheckpoint received from primary. This is problematic as primary performs validations on this received replication checkpoint from replica shard (e.g. recently we started comparing codec names for bwc).

Expected behavior

1. During force segment sync, the replica/target should send the codec defined in engine config.
2. Send replication checkpoint derived from replica shard.

Thanks @mch2 for figuring out