Spider middlewares are not async compatible

[VMRuiz]

This library defines several middlewares that are added by default to all jobs running on Scrapycloud. People wanting to use async spider middlewares may face issues as described here

My proposal is to add async compatibility by implementing process_spider_output_async method as per docs to avoid such issues while keeping compatibility with previous versions. Alternatively, they could be converted into fully async middlewares limiting the futures releases of the library to Scrapy >=2.7