hwlat: Add per cpu mask for hwlat_detector

Instead of just running a single kernel thread to do the hardware latency
detection, run one on each CPU that the user specificies. By default
this will be just one CPU (on CPU 0). But the user may specify what
CPUs they would like to run the tests on.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
5 files changed