Readless increments are not used, even when enabled

Use of readless increments is currently disabled by default for table datasets. However, even when enabled by configuration, readless increment operations are still not used. When a dataset is configured with dataset.table.readless.increment=true, the IncrementHandler coprocessor is configured on the table, but calling BuffereingOrderedTable.increment() still goes through the BufferingOrderedTable.incrementAndGet() path, so the operations are not performed as readless increments.