When compiled with 'options RSS', the ena driver created taskqueues
using taskqeueue_start_threads_cpuset passing a mask value of NULL,
in both ena_setup_tx_resources (for the enqueue path) and
ena_create_io_queues (for the completion-processing path).
In the default configuration, on most EC2 instances, this results in
taskqueues running in the right NUMA domain, but only by accident.
Set (struct ena_que)->domain and use that to inform the choice of CPU
sets. On a c8gn.48xlarge EC2 instance this doubles throughput on a
32-TCP-stream benchmark.
MFC after: 7 days
Sponsored by: Amazon