Description of problem:
Reshape of a RAID volume is broken after trying to stop all MD devices.

Version-Release number of selected component (if applicable):
- mdadm-3.2.3-6.fc17.x86_64

How reproducible:
Always

Steps to Reproduce:
1. Create an IMSM container and a volume with a spare:
# mdadm -CR /dev/md/imsm -e imsm -n 4 /dev/sda /dev/sdb /dev/sdc /dev/sdd
# mdadm -CR /dev/md/vol0 -l0 --size 2G -n3 /dev/sda /dev/sdb /dev/sdc

2. Run OLCE (or migration):
# export MDADM_EXPERIMENTAL=1
# mdadm -G /dev/md127 --raid-devices=4

3. Wait until reshape starts.
4. Try to stop all MD devices:
# mdadm -Ss

Actual results:
MD devices are not stopped. Reshape seems to be finished (metadata is in normal/clean state) but it is not finished, it is broken. Data are corrupted.

 mdadm: stopped /dev/md126
 mdadm: Cannot get exclusive access to /dev/md127:Perhaps a running process, mounted filesystem or active volume group?
 mdadm: Cannot get exclusive access to /dev/md127:Perhaps a running process, mounted filesystem or active volume group?
 mdadm: Cannot get exclusive access to /dev/md127:Perhaps a running process, mounted filesystem or active volume group?
 mdadm: /dev/md/vol0: could not set level to raid0

# cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4] [raid0]
md126 : active raid0 sda[3] sdb[2] sdc[1] sdd[0]
        8388608 blocks super external:/md127/0 128k chunks
md127 : inactive sda[3](S) sdb[2](S) sdc[1](S) sdd[0](S)
        4420 blocks super external:imsm
unused devices: <none>

Expected results:
Reshape should be stopped correctly on actual checkpoint.