This change persists into the offset record weather or not this partition originally marked itself as ready to serve.  This is then used on restart to determine if a node should report completion or not when coming online.

The intention is that when a node is lagged under normal conditions (things like heavy load or some other bad condition), we can still restart the node or do deployments and have the node come up and serve traffic.  This is done under the notion that it's generally preferable to be online and serving and stale as opposed to caught up/catching up, but unavailable.

This will not impact buffer replay on a version push because a replay won't pass a ready to serve check.