vllm - 💡(How to fix) Fix [Doc]: Mention Ascend NPU in the main quickstart

Problem

The main vLLM quickstart currently lists installation paths for NVIDIA CUDA, AMD ROCm, and Google TPU, but it does not mention Ascend NPU / vLLM Ascend.

For users who have Ascend NPUs, or users who arrive at the main vLLM quickstart while looking for non-CUDA accelerator instructions, this makes the supported path harder to discover. The Ascend project already has a dedicated quick start at:

https://docs.vllm.ai/projects/ascend/en/latest/quick_start.html

Suggested improvement

Add an "Ascend NPU" installation tab or note to docs/getting_started/quickstart.md, similar to the existing Google TPU entry, that points users to the vLLM Ascend quick start.

The main quickstart should probably avoid duplicating the full Ascend setup because the Ascend instructions depend on NPU hardware, driver/CANN versions, and container images. A short pointer would be enough to make the supported path discoverable.

Proposed validation

This is a documentation-only change. Validation can include:

• git diff --check
• documentation formatting / link validation if available

References

• Main quickstart: https://github.com/vllm-project/vllm/blob/main/docs/getting_started/quickstart.md
• vLLM Ascend quick start: https://docs.vllm.ai/projects/ascend/en/latest/quick_start.html