[Remote store] Replication handling for remote translog for primary & replicas

[ashking94]

Is your feature request related to a problem? Please describe.
Segment replication along with remote store for segments and translog makes creation of translog optional on replicas since translog is added for durability on the replicas. Now if we store the translog on remote store (like s3) durably, we can eliminate the need for translog on the replicas. On shard failures or failovers, the translog can be replayed from remote store. Irrespective of what approach we finalise in #3706, we need to implement a new engine for either no op replication or not supporting index/update/delete altogether. Since there would be changes in how checkpoints are perceived, we would also need to handle how the global checkpoint and local checkpoints are handled on primary and replica both. We also need to ensure that peer recovery and failovers are happening as currently.

Describe the solution you'd like
Creation of new engine that would run on replica especially. We would try to avoid making changes to Internal Engine which would run on the primary shard. This requires further brainstorming.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.