Currently, mergeAllChunks computes the set of chunks to merge while holding the critical section. This unnecessarily increases the time spent in the critical section. The logic that determines which chunks to merge should be moved outside the critical section, leaving only the work that must be performed under it.

This task should also include:

• adding a cluster parameter to limit the maximum number of chunks that can be merged in a single operation limited to maximum 100 chunks per mergeAllChunks.

• making the config server commit command for mergeAllChunks idempotent