docs: fix incorrect PEFT anchor link in fine-tuning section (#47927)

The PEFT Docs link in the bitsandbytes fine-tuning section pointed to
#aqlm-quantization instead of #quantize-a-model, sending readers to the
AQLM section instead of the relevant bitsandbytes/QLoRA content.