fix: Call wipefs -a in LonghornV2Provisioner.Format() (backport #157)

[mergify]

Problem:
When using the Longhorn V2 data engine, for NVMe devices where Longhorn decides to use the nvme bdev driver, device activation will fail if there's an existing filesystem. This does not appear to be a problem when using the aio driver.

Solution:
Run wipefs -a $devPath during provisioning.

Without this, we'll potentially have devices stuck unschedulable with an obscure error message like this:

Disk c1fd02d173a3f4b9176705e9b2e37d39(0003:03:00.0) on node altra is not ready: failed to generate disk config: error: rpc error: code = Internal desc = rpc error: code = Internal desc = failed to add disk block device: failed to create disk bdev: failed to attach NVMe disk 0003:03:00.0: error sending message, id 3126, method bdev_nvme_attach_controller, params {c1fd02d173a3f4b9176705e9b2e37d39 {PCIe 0003:03:00.0 } 30 2 15 disable}: {"code": -19,"message": "No such device"}

Related Issue:
harvester/harvester#6828

Test plan:

• On a host with an extra nvme device, run mkfs.ext4 /dev/nvme0n1 (or whatever the device name is) to make a filesystem.
• Try to add the disk to the host for Longhorn V2 usage. It should be stuck unschedulable (check kubectl -n longhorn-system get lhn -o yaml and look under diskStatus for an error message similar to the description above).
• Re-test with this fix applied. The disk should provision fine.

---

This is an automatic backport of pull request #157 done by [Mergify](https://mergify.com).