_compare_rows_by_primary_key_streaming 并非流式，还是全量缓存了

[cobolbaby]

diffTableTool/table_diff.py

Lines 1050 to 1066 in 38a6869

	for row in cursor1:
	row_dict = dict(zip(comparison_fields, row))
	key = tuple(row_dict[pk] for pk in primary_keys)
	rows1_data[key] = row_dict
	row_count1 += 1
	# 读取第二个表的数据
	row_count2 = 0
	for row in cursor2:
	row_dict = dict(zip(comparison_fields, row))
	key = tuple(row_dict[pk] for pk in primary_keys)
	rows2_data[key] = row_dict
	row_count2 += 1
	# 收集所有主键值
	all_keys = set(rows1_data.keys()) | set(rows2_data.keys())
	logger.info(f"总共 {len(all_keys)} 个唯一主键值")

1kw+ 的表对比比较吃力，太吃内存资源