fix: set LHv2 DiskDriver to "auto" if unset

[tserong]

Problem:
We need to force DiskDriver to "auto" if it's not explicitly set, because Longhorn also does that internally. If we don't do this, the subsequent reflect.DeepEqual() in LonghornV2Provisioner.Update() will always fail because we have an empty string, but the LHN CR will have it set to "auto" which results in a weird resync loop.

Solution:
This PR.

Related Issue:
harvester/harvester#6709

Test plan:
Add an LHv2 disk. Verify we don't see the errors mentioned in the related issue.

[tserong]

harvester/vagrant-k3s#3 should fix that CI failure

[Vicente-Cheng]

LGTM, thanks!

[Vicente-Cheng]

@Mergifyio backport v0.7.x

[mergify]

backport v0.7.x

✅ Backports have been created

• #154 fix: set LHv2 DiskDriver to "auto" if unset (backport #152) has been created for branch v0.7.x

[tserong]

latest CI failure is:

=== RUN   TestSingleDiskOperation/Test_1_UnprovisionSingleDisk
    test_0_single_disk_test.go:141: 
        	Error Trace:	/home/github/actions-runner/_work/node-disk-manager/node-disk-manager/tests/integration/test_0_single_disk_test.go:141
        	Error:      	Not equal: 
        	            	expected: "/var/lib/harvester/extra-disks/9665fd40086d48e7ca541b1e990c6645"
        	            	actual  : ""
        	            	
        	            	Diff:
        	            	--- Expected
        	            	+++ Actual
        	            	@@ -1 +1 @@
        	            	-/var/lib/harvester/extra-disks/9665fd40086d48e7ca541b1e990c6645
        	            	+
        	Test:       	TestSingleDiskOperation/Test_1_UnprovisionSingleDisk
        	Messages:   	Mountpoint should be empty after we remove disk!

I suspect this is probably because Test_1_UnprovisionSingleDisk() is still setting Spec.FileSystem.Provisioned = false rather than Spec.Provision = false, and is unrelated to this PR.

[Vicente-Cheng]

Yeah, and it should be fixed on #151. I thought we could merge it and let CI re-run with the master branch