disk_stress_ng test doesn't always find the NVME largest partition

[Jeff Lane ]

In this test result:
https://certification.canonical.com/hardware/201711-25906/submission/138996/test/70803/result/10413455/

the disk_stress_ng test finds the NVME device and knows that nvme0n1p1 is a partition, but then, for some inexplicable reason declares it doesn't exist.

HOWEVER, after re=running that test, the disk_stress_ng test finally finds the partition and mounts it with no intervention necessary (so basically just re-running the test a couple times finally works).

Here's the same system with a passing result after simply re-running the test a couple times:
https://certification.canonical.com/hardware/201711-25906/submission/138997/test/70803/result/10413510/

So initially, at least, this looks like some sort of race condition where we are or are not getting the partition data in time.

Need to investigate this and see if we can recreate it and fix whatever is triggering this.