Fix model offload bug when key isn't present

[DN6]

What does this PR do?

This PR fixes an issue with enable_model_cpu_offload when a key isn't present in a checkpoint on the hub

Fixes # (issue)
#5026

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?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[patrickvonplaten]

LGTM, the use case here is the refiner mostly: https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0 which doesn't have a text_encoder. We really need tests for this whole pipeline class - opening an issue about it.

[patrickvonplaten]

Related: #5031

[DN6]

There's a flaky test. Will merge when CI is green.

[tin2tin]

There's a flaky test. Will merge when CI is green.

Thank you. I'm standing by to release the Pallaidium Blender add-on with Würstchen implented, when this fix is comitted.

[tin2tin]

Thank you for the swift handling of this.