Make coverstore scalable about openlibrary

[anandology]

Noticed that the size of small and medium size covers is less than 100GB. One good way to reduce the load is by sending the large and original image traffic to petabox. If the covers are stored in zip files instead of tar files like we do now, zipview.php can be used to serve individual images.

Also smaller sized items are easier to handle on archive.org. I think we should create one item for each 10K images, with one zip file for each size. In this new approach item olcovers-42 will have:

• olcovers-42.zip - original covers from 420000.jpg to 429999.jpg
• olcovers-42-s.zip - small sized covers from 420000-S.jpg to 429999-S.jpg
• olcovers-42-m.zip - medium sized covers from 420000-M.jpg to 429999-M.jpg
• olcovers-42-l.zip - large sized covers from 420000-L.jpg to 429999-L.jpg
• olcovers-42-json.zip - metadata about the covers from 420000.json to 429999.json

[anandology]

The metadata for each cover contains:

{
    "id": 1,
    "olid": "OL9209023M",
    "author": "/people/anand",
    "source_url": "http://www.example.com/foo.jpg",
    "width": 200,
    "height": 500,
    "timestamp": "2007-12-31T06:47:00.010207" 
 }

[anandology]

Couldn't get time to work on this in July. Moving to August milestone.

[anandology]

Looks like we are not going to work on this soon.

Adding "On Hold" label and closing the bug. We'll reopen if we want to work on it.