Make SD3Transformer2DModel hidden states contiguous

[menglcai]

What does this PR do?

Ensures hidden states are contiguous after each JointTransformerBlock call in SD3Transformer2DModel.

On ROCm, non-contiguous tensors produced by JointTransformerBlock cause performance degradation that accumulates across transformer blocks.

Benchmarks (SD3-medium, 512×512, 28 steps, fp16):

Platform	PyTorch	Baseline	Patched	Speedup
AMD Radeon(TM) 8060S Graphics (gfx1151)	2.12.0+rocm7.15.0a20260711	8503 ms	7470 ms	+12.1%
AMD RX 9070 XT (gfx1201)	2.12.0+rocm7.15.0a20260713	2547 ms	2358 ms	+7.4%
NVIDIA RTX 5090	2.12.1+cu132	1042 ms	1017 ms	within noise

Before submitting

• Did you use an AI agent (Claude Code, Codex, Cursor, etc.) to help with this PR? If so:
  • Did you read the Coding with AI agents guide?
  • Did you self-review the diff against .ai/review-rules.md?
• 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?
• Are you the author (or part of the team) of the model/pipeline (only applicable for model/pipeline related PRs)?

Who can review?

@sayakpaul @yiyixuxu

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.

[sayakpaul]

Could also comment on the position in the block of code where contiguous() should be applied?

[menglcai]

You're right. After investigating, the non-contiguous tensor originates from PatchEmbed.forward - flatten(2).transpose(1, 2) (BCHW → BNC) produces a non-contiguous layout.

diffusers/src/diffusers/models/embeddings.py

Line 562 in 4f37263

latent = latent.flatten(2).transpose(1, 2) # BCHW -> BNC

So the fix is more precisely placed right after self.pos_embed(), not inside the loop.

Updated the patch: moved .contiguous() to immediately after self.pos_embed(hidden_states), dropped the encoder_hidden_states call, and added a comment explaining the root cause.