Introduce dedicated OpCounters-like metrics for resharding and use them in ReshardingOplogApplier

ReshardingOplogApplier currently increments globalOpCounters when applying inserts, updates, and deletes. Using globalOpCounters can make it difficult to distinguish in the FTDC/t2 output the ops/sec from the client workload from the ops/sec from resharding as it tries to keep up with the client writes.

Having a new shardingStatistics.resharding.opcounters subsection would enable splitting up these metrics in ReshardingOplogApplier from the globalOpCounters.

shardingStatistics:

  ... (other existing metrics) ...

  resharding:

    ... (other existing metrics) ...

    opcounters:

      insert: int64

      update: int64

      delete: int64