sched: EAS: add infrastructure for core_ctl

- Add sched_get_cpu_util() API that returns the CPU utilization
in pct for the last window tracked by WALT. core_ctl can use
this API to query the load of all CPUs from WALT window rollover
callback.

- Track number of misfit tasks (also called big tasks). A task
is considered misfit when its utilization is more than the
capacity of the CPU that is hosting it. misfit flag is added
per task so that we don't have to deal with updating
rq->hmp_stats.nr_big_tasks when task boost is changed or cpufreq
policy is updated. When a task is running on max capacity, it
won't be called misfit task. Any task running on the max capacity
CPU is considered as a BIG task by sched_get_nr_running_avg() API.

Change-Id: Ia15962f346e52bd3cfa08bb6f119082840ca893d
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>