[Test] Fix test_patching_cluster false failures on Lustre modules and private-OS AMI drift

[gmarciani]

Description of changes

Fix test_patching_cluster false failures on Lustre modules and private-OS AMI drift

In particular:

1. The FSx Lustre mount is on-demand (x-systemd.automount), so its client modules aren't loaded on the head node right after the patch reboot. Trigger the mount before the post-patch snapshot and define the mount dir once in the test, injecting it into the cluster config.

2. Pin private OSes (rocky8/rocky9) to the AMI the cluster was created with on updates, since the framework re-injects the latest private AMI on every render and the update would otherwise drift to a newer, possibly unavailable, AMI.

Also added a log line to report the active kernel after the patching to facilitate troubleshooting.

Tests

SUCCESS:

test-suites:
  patching:
    test_patching.py::test_patching_cluster:
      dimensions:
        - regions: [{{ g4dn_8xlarge_CAPACITY_RESERVATION_3_INSTANCES_2_HOURS_NOPG_rocky8 }}]
          instances: ["g4dn.8xlarge"]
          oss: ["rocky8"]
          schedulers: ["slurm"]

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.