sched: Fix lockdep warning for rq locks

WALT tries to acquire all possible cpu's runqueue locks to update
cluster stats or capacities. When lockdep enabled, it reports below
warning.

Fix it by using raw_spinlock_nested API to inform lockdep it is
intentional.

Report:

[1.365691] May be due to missing lock nesting notation\x0a
[1.372528] 1 lock held by swapper/0/1:
[1.376399] #0:  (&rq->lock){-.-...}, at: [<ffffff98605282b0>]
    	acquire_rq_locks_irqsave.constprop.22+0x64/0x9c
[1.386649] \x0astack backtrace:
[1.391054] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.9.106+ #10
[1.397282] Hardware name: Qualcomm Technologies, Inc. SDM845 v2.1 MTP (DT)
[1.404295] Call trace:
[1.406769] [<ffffff986048c2bc>] dump_backtrace+0x0/0x2c0
[1.412217] [<ffffff986048c59c>] show_stack+0x20/0x28
[1.417315] [<ffffff98608ba190>] dump_stack+0xa8/0xe0
[1.422412] [<ffffff9860542118>] __lock_acquire+0x18fc/0x1d74
[1.428204] [<ffffff9860542964>] lock_acquire+0xe0/0x288
[1.433569] [<ffffff986172263c>] _raw_spin_lock+0x44/0x58
[1.439009] [<ffffff98605282b0>] acquire_rq_locks_irqsave.constprop.22+0x64/0x9c
[1.446458] [<ffffff9860528324>] update_all_clusters_stats+0x3c/0x1f8
[1.452953] [<ffffff986052c494>] update_cluster_topology+0x2a0/0x300
[1.459363] [<ffffff9861fc456c>] sched_init_smp+0xc4/0x11c
[1.464896] [<ffffff9861fb0e48>] kernel_init_freeable+0x150/0x2bc
[1.471044] [<ffffff98617198d4>] kernel_init+0x18/0x100
[1.476308] [<ffffff9860483f00>] ret_from_fork+0x10/0x50

Change-Id: Ibc5726e0f50f0135fd9b46c8065ee0f52dd5d5e7
Signed-off-by: Lingutla Chandrasekhar <clingutla@codeaurora.org>