We should remove the ASYNC_OPERATIONS_EXECUTOR and PERSISTENCE_EXECUTOR components and replace them with a NON_BLOCKING_EXECUTOR and BLOCKING_EXECUTOR or something similarly named.

We should probably default the NON_BLOCKING_EXECUTOR to 2 * numCPU and blocking to 100? The queue size should be a bit bigger for blocking as compared to non blocking. The parser should allow these to be changed as other executors are currently.