* increase iterations to 10K this produced more stable averages
* add options to dump all measurements
* add baseline benchmark using two threads and read/write syscalls
* add iouring benchmark using two threads and iouring to read