test: improve flaky usb reconnection test

[yaroslavborbat]

Description

Refactor the flaky USB reconnection e2e test to improve reliability.

The monolithic mountUSB() function was split into three focused steps:

• findUSBMountDevice() — locates the block device path via /sys/bus/usb/devices with retries
• formatUSBDevice(mountDevice) — formats the device with mkfs.vfat before the first mount to ensure a clean filesystem
• mountUSBDevice(mountDevice) — mounts the located device directly by path (no multi-fallback mount -t auto)

Additional diagnostic GinkgoWriter.Println calls were added at key steps, and LongTimeout is now used for the mount step to handle slow environments.

Why do we need it, and what problem does it solve?

The USB reconnection test was flaky because:

1. Device lookup and mounting were bundled in one shell script, making it hard to diagnose which step failed.
2. The mount command tried several fallback options (-t auto, -t vfat, no flag), masking root causes.
3. The device was not formatted before mounting, so an unclean or unformatted device could cause a mount failure on first use.

Splitting the steps, adding an explicit mkfs.vfat format, and increasing the mount timeout makes the test deterministic and easier to debug when it does fail.

What is the expected result?

1. Run the VirtualMachineUSB e2e suite.
2. The test should consistently pass without flaky failures on the "Mounting USB device" and "Remounting USB device after migration" steps.
3. On failure, GinkgoWriter output clearly shows which step (find / format / mount) failed.

Checklist

• The code is covered by unit tests.
• e2e tests passed.
• Documentation updated according to the changes.
• Changes were tested in the Kubernetes cluster manually.

Changelog entries

section: test
type: fix
summary: Improve reliability of flaky USB reconnection e2e test by splitting mount logic into find, format, and mount steps.
impact_level: low