Fix issues when running multiple LoRA tests on the v6e-8 machine.

[vanbasten23]

Description

Problem: When I run my CI test https://github.com/vllm-project/tpu-inference/blob/main/tests/lora/test_lora.py on the v6e-8 CI machine, the first test succeeds and the second one fails with the error jaxlib._jax.XlaRuntimeError: UNKNOWN: TPU initialization failed: open(/dev/vfio/1): Device or resource busy: Device or resource busy; Couldn't open iommu group /dev/vfio/1. It appears that after the first test finishes, it didn't release the TPU device. It's persistent. In my local env, ~70% of the chance I could reproduce the issue. Also I couldn't reproduce the issue on the v6e-1 machine in the CI.

This PR intends to fix the above problem.

Tests

CI

Checklist

Before submitting this PR, please make sure:

• I have performed a self-review of my code.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have made or will make corresponding changes to any relevant documentation.

[github-actions]

Description

Start with a short description of what the PR does and how this is a change from
the past.

The rest of the description includes relevant details and context, examples:

• why is this change being made,
• the problem being solved and any relevant context,
• why this is a good solution,
• some information about the specific implementation,
• shortcomings of the solution and possible future improvements.

If the change fixes a bug or a Github issue, please include a link, e.g.,:
FIXES: b/123456
FIXES: #123456

Tests

Please describe how you tested this change, and include any instructions and/or
commands to reproduce.

Checklist

Before submitting this PR, please make sure:

• I have performed a self-review of my code.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have made or will make corresponding changes to any relevant documentation.