Add unzip_above option

[dadoonet]

Based on #690 (comment) discussion, we can implement a new unzip_above: 100mb option which would extract in a tmp dir (defined by a FSCRAWLER_TMP system option (default to System tmp dir)) and run then FSCrawler on that dir.

For each generated Doc, we would then transform the URL as coming not from the tmp dir but from the ZIP/GZ file.

Then after run, we would remove the tmp dir and files. Could be optional with a setting like keep_tmp_files: true.

[dadoonet]

So this is already happening behind the scene in Tika. No need for that then.