Update gaudi-docker to v1.24.0 & fix CUDA UT

[XuehaoSun]

Description

This pull request updates dependencies and improves test script robustness in the CI pipeline. The most significant changes are upgrades to key package versions, improved handling of test duration extraction, and changes to how pre-release dependencies are installed.

Dependency and environment updates:

• Upgraded gptqmodel and its requirements from version 5.8.0 to 6.0.3 in .azure-pipelines/scripts/cuda_unit_test/run_cuda_ut.sh to use the latest release from the ModelCloud/GPTQModel repository.
• Updated the sglang dependency in test/test_cuda/requirements_sglang.txt from a direct GitHub reference to a PyPI version requirement (sglang >= 0.5.10).
• Updated the Habana docker image in .azure-pipelines/template/docker-template.yml from 1.23.0/pytorch-installer-2.9.0 to 1.24.0/pytorch-installer-2.10.0 for both pull and run steps. [1] [2]

Testing and installation improvements:

• Added the --prerelease=allow flag to uv pip install commands for requirements_sglang.txt in both CUDA unit test scripts to allow installation of pre-release package versions. [1] [2]

Test result parsing:

• Improved duration extraction in .azure-pipelines/scripts/ut/collect_result.py to use the last matched time pattern in the test output, ensuring the most accurate test duration is reported.

Type of Change

• Bug fix
• New feature
• Documentation update
• Performance improvement
• Code refactoring
• Other (please specify):

Related Issues

Fixes or relates to #

Checklist Before Submitting

• My code has been tested locally.
• Documentation has been updated as needed.
• New or updated tests are included where applicable.

[Copilot]

Pull request overview

Updates the CI pipeline dependencies used for Habana Gaudi and CUDA unit test runs, and improves duration parsing for test log summaries.

Changes:

• Bump Habana Gaudi Docker image reference from gaudi-docker/1.23.0 (PyTorch installer 2.9.0) to gaudi-docker/1.24.0 (PyTorch installer 2.10.0) in the Azure Pipelines docker template.
• Adjust log duration extraction to use the last reported in <seconds>s occurrence in log tail content.
• Update CUDA unit test dependency install to GPTQModel v6.0.3 (wheel + requirements).

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
.azure-pipelines/template/docker-template.yml	Updates the pulled/run Habana Gaudi container image version used in CI.
.azure-pipelines/scripts/ut/collect_result.py	Improves duration parsing robustness by selecting the last time match in logs.
.azure-pipelines/scripts/cuda_unit_test/run_cuda_ut.sh	Bumps GPTQModel dependency version used in CUDA unit test setup.

[XuehaoSun]

/azp run Unit-Test-CUDA-AutoRound

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).