Force to refresh datastore cache for listVMDK.

[lipingxue]

This is the first step for the fix of #818.
In this change, we force to refresh datastore cache for listVMDK, which provides a workaround for #818.

[lipingxue]

This is the test to verify this fix.
Step1: run "docker volume ls" to list volumes

root@photon-eZ5ILvREQ [ ~ ]# docker volume ls
DRIVER              VOLUME NAME
vmdk                vol1@new-datastore1
vmdk                vol2@new-datastore1
vmdk                vol3@new-datastore1
vmdk                vol4@new-datastore1
vmdk                vol5@new-datastore1
vmdk                vol6@new-datastore1
vmdk                vol7@new-datastore1

Step2: change the datastore name from "new-datastore1" to "datastore1"

Step3: try to create a new volume, which failed

root@photon-eZ5ILvREQ [ ~ ]# docker volume create --driver=vmdk --name=vol8
Error response from daemon: create vol8: VolumeDriver.Create: Server returned an error: IndexError('list index out of range',)

Step4: run "docker volume ls", which list the volume with new datastore name(correct)

root@photon-eZ5ILvREQ [ ~ ]# docker volume ls
DRIVER              VOLUME NAME
vmdk                vol1@datastore1
vmdk                vol2@datastore1
vmdk                vol3@datastore1
vmdk                vol4@datastore1
vmdk                vol5@datastore1
vmdk                vol6@datastore1
vmdk                vol7@datastore1

Step5: try to create a new volume, which succeeded

root@photon-eZ5ILvREQ [ ~ ]# docker volume create --driver=vmdk --name=vol8
vol8
root@photon-eZ5ILvREQ [ ~ ]# docker volume ls
DRIVER              VOLUME NAME
vmdk                vol1@datastore1
vmdk                vol2@datastore1
vmdk                vol3@datastore1
vmdk                vol4@datastore1
vmdk                vol5@datastore1
vmdk                vol6@datastore1
vmdk                vol7@datastore1
vmdk                vol8@datastore1

[lipingxue]

Please review this change @msterin @shaominchen

[shaominchen]

LGTM.