Zero-Shot Cross-Lingual Reranking with Large Language Models for Low-Resource Languages