[Fix] fix device orientation for image processor

[ZailiWang]

Motivation

In BaseMultimodalProcessor::process_mm_data, the device was set as cuda or npu in a previous PR, leading to error for other backends.

Modifications

Corrected for cpu backend. Other backends may need to do the same to enable the image processor.
The installation of torchaudio is piggy-backed, as latest SGL would require it for multimodal execution.

Accuracy Tests

N/A

Benchmarking and Profiling

N/A

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.
• Work with maintainers to merge your PR. See the PR Merge Process

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[ZailiWang]

/tag-run-ci-label

[mingfeima]

@blzheng take a look.

[JustinTong0323]

/tag-and-rerun-ci

[ZailiWang]

Hi @JustinTong0323 would you please help merge the PR? I have resolved the conflicts.

[ZailiWang]

The CI failures are all timeouts and should be irrelevant to the change in this PR.

[ZailiWang]

Hi @JustinTong0323 would you help merge the PR? Thanks.