[Advanced Diffusion Training] Adapt train_dreambooth_lora_sdxl_advanced.py to use peft #6118

With #5388 we adapted our training examples that make use of LoRA to use peft. It'd be very nice to do the same for advaced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py.

Unable to assign @linoytsaban for some reason.

Cc: @younesbelkada for visibility.