Currently timeouts and retry strategies can be applied at the level of the underlying SDK client, but this is near-global (it's possible to have different clients for snapshot, journal, and projection/query, but even then the same timeout would apply for (e.g.) reading and writing snapshots). Does it make sense to have retries at the plugin level where a per-operation strategy can be deployed?
Currently timeouts and retry strategies can be applied at the level of the underlying SDK client, but this is near-global (it's possible to have different clients for snapshot, journal, and projection/query, but even then the same timeout would apply for (e.g.) reading and writing snapshots). Does it make sense to have retries at the plugin level where a per-operation strategy can be deployed?