msm: ssr: remove msleep in the panic handler

msleep in the panic handler causes another panic.

bad: scheduling from the idle thread!
[<c00144ac>] (unwind_backtrace+0x0/0x11c)
[<c00a4a08>] (dequeue_task_idle+0x18/0x28)
[<c009f864>] (dequeue_task+0xb0/0xc0)
[<c08294d4>] (__schedule+0xfc/0x6d4)
[<c0828294>] (schedule_timeout+0x2fc/0x358)
[<c0085804>] (msleep+0x14/0x20)
[<c0057368>] (ssr_panic_handler+0x24/0x44)
[<c082cf90>] (notifier_call_chain+0x38/0x68)
[<c082cff4>] (__atomic_notifier_call_chain+0x34/0x44)
[<c082d018>] (atomic_notifier_call_chain+0x14/0x18)
[<c081e814>] (panic+0xb0/0x1d8)
[<c0011bb4>] (die+0x1d0/0x218)
[<c081e498>] (__do_kernel_fault.part.8+0x54/0x74)
[<c082ce5c>] (do_page_fault+0x39c/0x3c0)
[<c0008524>] (do_PrefetchAbort+0x34/0x98)

this panic loop is not terminated until watchdog timer expiration.
msleep is replaced by mdelay. it should be checked if this 3 seconds
delay is reqired or not.