fastfat: Pool corruption after unexpected device deletion (x64)

This is what happens:

• usetup formats the volume
• usetup closes the volume (calls NtClose)
• the kernel closes the handle and calls IopCloseFile
• IopCloseFile sends an IRP to vfat, calling VfatCleanup
• VfatCleanup acquires the resource DeviceExt->DirResource
• VfatCleanup calls VfatCleanupFile
• VfatCleanupFile calls VfatCheckForDismount
• VfatCheckForDismount calls IoDeleteDevice(DeviceExt->VolumeDevice);
• IoDeleteDevice calls IopUnloadDevice
• IopUnloadDevice deletes the device and frees the device extension
• back to VfatCleanup, the resource is released, but it has already been freed

The issue could be related to the "UnCleanCount hack in VfatCheckForDismount, where a reference count of 3 prevents the device from being deleted.