diffusers docs have too many `cuda` hardcoding, this is unfriendly to device-agnostic

[kaixuanliu]

Describe the bug

As title

Reproduction

Some examples

- https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/pipelines/animatediff.md#animatediffcontrolnetpipeline
- https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/pipelines/anyflow.md#generation-with-anyflow-bidirectional-t2v
- https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/pipelines/cosmos.md#basic-usage

Logs

System Info

- 🤗 Diffusers version: 0.40.0.dev0
- Platform: Linux-6.17.0-41-generic-x86_64-with-glibc2.39
- Running on Google Colab?: No
- Python version: 3.12.3
- PyTorch version (GPU?): 2.13.0+xpu (False)
- Huggingface_hub version: 1.26.1
- Transformers version: 5.16.0.dev0
- Accelerate version: 1.14.0
- PEFT version: 0.19.1
- bitsandbytes version: 0.49.2
- gguf version: 0.19.0
- optimum-quanto version: 0.2.7
- torchao version: 0.17.0
- Safetensors version: 0.8.0
- xFormers version: not installed
- Accelerator: NA
- Using GPU in script?: YES
- Using distributed or parallel set-up in script?: NO

Who can help?

@stevhliu