block: update nsecs[] in part_stat_show() and diskstats_show()

hulk inclusion
category: bugfix
bugzilla: https://gitee.com/openeuler/kernel/issues/I57S8D


CVE: NA

--------------------------------

commit 7ec2ec68 ("block: update io_ticks when io hang") fixed that
%util will be zero for iostat when io is hanged, however, avgqu-sz is
still zero while it represents the number of io that are hunged. On the
other hand, for some slow device, if an io is started before and done
after diskstats is read, the avgqu-sz will be miscalculated.

To fix the problem, update 'nsecs[]' when part_stat_show() or
diskstats_show() is called. In order to do that, add 'stat_time' in
struct hd_struct and 'rq_stat_time' in struct request to record the
time. And during iteration, update 'nsecs[]' for each inflight request.

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com>