[Remote Store] Durability issue with indices created via split/shrink/clone/manual snapshot recovery

[gbbafna]

Describe the bug

For remote store backed indices, if a user does split/shrink/clone/manual snapshot recovery, we don't guarantee that the segments created gets uploaded to remote, before taking in new writes.

In absence of this check, we won't be able to durable persist the index , by just using the Remote Translog for newer writes . This is because Remote Segment Store in combination of Remote Translog won't be able to construct the whole data via remote restore api.

Expected behavior

Before we enable writes on indices created via split/shrink/clone/manual snapshot recovery, the segments should be durably persisted.

This makes sure that at any point of time, the data is durably persisted.