sched/deadline: Return best-fit CPUs in global CPU DL mgmt for asymmetric CPU capacities

Instead of returning all fitting CPUs for a task, only return those with
the minimal CPU (original) capacity.

This potentially leaves CPUs with higher (original) capacity for more
demanding (i.e. higher dl_density) tasks.

Migrations due to the fact that there won't be any fitting free CPUs
(from DL perspective) available the moment those tasks get enqueued
could be avoided this way.

This patch reworks the idea initially presented in:

https://lkml.kernel.org/r/20190506044836.2914-7-luca.abeni@santannapisa.it



Signed-off-by: Dietmar Eggemann <dietmar.eggemann@arm.com>