[lora training] add caption dropout and aspect ratio buckets to krea2 and qwen-image

[linoytsaban]

Adds aspect-ratio bucketing and caption dropout to the krea2 and qwen-image dreambooth LoRA scripts (neither supported these before):

• Aspect-ratio bucketing for krea2 and qwen-image.
• Caption dropout (--caption_dropout): randomly replace an instance caption with "" so that fraction trains unconditionally (improves CFG).
• --bucket_no_upscale: pad images smaller than their bucket instead of upscaling; class/prior-preservation images are bucketed too.

Split from the original PR; bucketing fixes to the other scripts that already support buckets are in #14158.

[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.

[linoytsaban]

krea comparison with and without buckets and dropout

[sayakpaul]

Should we do it just one and then propagate to the others? This way, it's easier to review.

[linoytsaban]

Should we do it just one and then propagate to the others? This way, it's easier to review.

I can split the krea 2 + qwen image changes (that introduce aspect ratio buckets + caption dropout since they dont currently support these at all) from the changes to the other scripts (that already support bucketing) where it's fixing bucketing issues. wdyt?

[sayakpaul]

Yes sure

[sayakpaul]

Same comments as the other PR. WDYT?