fhir-bucket client-driven reindex retry should try smaller chunks each time

[punktilious]

Is your feature request related to a problem? Please describe. If the reindex operation request fails due to repeated deadlocks, the client will retry the request.

Describe the solution you'd like The client should reduce the number of resources (by half) for each retry. This should help to reduce the probability of hitting a deadlock repeatedly.

Describe alternatives you've considered Increase the number of retries.

Acceptance Criteria

Additional context N/A