Breaks on long filenames

[jonasob]

Requesting 50 files at once generally work, but not if the file names happen to be long. In this case the request fails since the URL is longer than ~2048 bytes, which seems to be about the longest that is supported. Would be good to figure out exactly where the limit is here (urllib?) and make sure that if the URL generated isn't longer than this - if 50 files doesn't fit, the request should probably be run twice or as many times as is needed to process all files in the task.