Create a scrapy backed cache based on WARC

[turicas]

If we extract the WARC write code from the spider and then create a WARC read routine we can implement a scrapy cache storage which reads and writes WARC files. This could be very handy to create archives from running old spiders just changing the cache setting.