drm/amdgpu: Optimize mutex usage (v4)

drm/amdgpu: Optimize mutex usage (v4)

In original function amdgpu_bo_list_get, the waiting
for result->lock can be quite long while mutex
bo_list_lock was holding. It can make other tasks
waiting for bo_list_lock for long period.

Secondly, this patch allows several tasks(readers of idr)
to proceed at the same time.

v2: use rcu and kref (Dave Airlie and Christian König)
v3: update v1 commit message (Michel Dänzer)
v4: rebase on upstream (Alex Deucher)

Signed-off-by: Alex Xie <AlexBin.Xie@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>