fix qwenimage mask ordering, tensor images, layered output, vae clamp, and exports

[akshan-main]

What does this PR do?

Part of #13581 (qwenimage review). Fixes six of the findings:

• Batched prompts with num_images_per_prompt > 1 paired attention masks with the wrong prompt: prompt_embeds expanded to [p0, p0, p1, p1] but the mask to [p0, p1, p0, p1]. Both now use repeat_interleave (the embeds tensor is unchanged, only the mask alignment is fixed).
• Tensor image inputs crashed on image.size (a method on Tensor, not a tuple) before preprocessing, in the edit / edit-inpaint / edit-plus / layered pipelines and the modular encoder. Now read the shape for tensors.
• QwenImageLayeredPipeline(output_type="latent") returned an undefined images (the latent branch assigned image), so it raised instead of returning latents.
• Layered zero_cond_t doubled timestep but not additional_t_cond, so batch > 1 mismatched in the time embedding. Now both are doubled.
• Tiled VAE decode skipped the [-1, 1] clamp that plain decode applies, so tiling shifted the output range. Now clamped, matching _decode and the Wan VAE.
• __init__ lazily exported QwenImagePriorReduxPipelineOutput and ReduxImageEncoder, neither of which exists here (Redux is a Flux feature), so importing the subpackage failed. Removed.

The guidance-path finding is a dead code path (nothing sets guidance_embeds), and the test-coverage finding is tracked separately as issue 8 of #13581.

Fixes #13581
Part of the huge #13656 take over

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? Discussed on Slack with @yiyixuxu
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

Who can review?

@yiyixuxu

[github-actions]

Hi @akshan-main, thanks for the PR! It does not appear to link an issue it fixes. If this PR addresses an existing issue, please add a closing keyword (e.g. Fixes #1234) to the PR description so the issue is linked. See the contribution guide for more details. If this PR intentionally does not fix a tracked issue, a maintainer can add the no-issue-needed label to silence this reminder.