ENH: Improve speed of function expanding LoRA scales

[BenjaminBossan]

What does this PR do?

Resolves #11816

The following call proved to be a bottleneck when setting a lot of LoRA adapters in diffusers:

diffusers/src/diffusers/loaders/peft.py

Line 482 in cdaf84a

weights = scale_expansion_fn(self, weights)

This is because we would repeatedly call unet.state_dict(), even though in the standard case, it is not necessary:

diffusers/src/diffusers/loaders/unet_loader_utils.py

Line 55 in cdaf84a

unet.state_dict(),

This PR fixes this by deferring this call, so that it is only run when it's necessary, not earlier.

Note: This PR doesn't change the fact that set_adapters becomes slower the more adapters are already loaded, but since it speeds up the whole process by a factor of approximately 10x, set_adapters is much less of a bottleneck.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[sayakpaul]

Thanks, LGTM!

Concerned tests:

• diffusers/tests/lora/utils.py

  Line 1108 in 05e7a85

  def test_simple_inference_with_text_denoiser_multi_adapter_block_lora(self):

• diffusers/tests/lora/utils.py

  Line 1182 in 05e7a85

  def test_simple_inference_with_text_denoiser_block_scale_for_all_dict_options(self):

[sayakpaul]

Since this is a private function, this should be more than okay to break

[sayakpaul]

Thanks for this contribution!