[Test] Add new e2e test use deepseek-v2-lite in ge graph mode

[ForBetterCodeNine]

What this PR does / why we need it?

The current test cases lack end-to-end (e2e) testing for the deepseek-v2-lite network in ge graph mode.

Does this PR introduce any user-facing change?

No

How was this patch tested?

• vLLM version: v0.11.0
• vLLM main: vllm-project/vllm@83f478b

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

This pull request adds end-to-end tests for the deepseek-v2-lite model in graph mode, which is a valuable addition for test coverage. The implementation is straightforward. I have one suggestion to refactor the newly added test functions using pytest.mark.parametrize to reduce code duplication and improve maintainability.

[gemini-code-assist]

The three new test functions test_e2e_deepseekv2lite_with_torchair, test_e2e_deepseekv2lite_with_torchair_ms_mla, and test_e2e_deepseekv2lite_with_torchair_v1scheduler are very similar and contain duplicated code, especially the additional_config dictionary which is identical in two of the tests. This makes the tests harder to maintain, as future changes might be missed in one of the copies.

To improve maintainability and reduce code duplication, you can refactor these three functions into a single parameterized test using pytest.mark.parametrize. This will make the test suite cleaner and easier to extend with more configurations in the future.

As a minor note, the helper function _deepseek_v2_lite_torchair_test_fixure has a typo and should be renamed to _deepseek_v2_lite_torchair_test_fixture. The suggestion below uses the corrected name, so you will also need to rename the function definition at line 231.

@pytest.mark.parametrize(
    "config_updates, use_v1_scheduler",
    [
        ({}, False),
        ({"enable_multistream_mla": True}, False),
        ({}, True),
    ],
    ids=[
        "default",
        "ms_mla",
        "v1scheduler",
    ])
def test_e2e_deepseekv2lite_with_torchair(config_updates, use_v1_scheduler):
    additional_config = {
        "torchair_graph_config": {
            "enabled": True,
        },
    }
    additional_config["torchair_graph_config"].update(config_updates)
    _deepseek_v2_lite_torchair_test_fixture(additional_config,
                                           use_v1_schduler=use_v1_scheduler)