Retry if a request fails

[sneakypete81]

Very occasionally API requests to the hosted site fail:

• When deleting a photo (https://travis-ci.org/photo/openphoto-python/builds/7605994):

  OpenPhotoError: Code 500: cURL resource: Resource id #15; cURL error: Unknown SSL protocol error in connection to awesomeness.openphoto.me.s3.amazonaws.com:443  (35)
  

• When uploading a photo (https://travis-ci.org/photo/openphoto-python/builds/7607189):

  OpenPhotoError: Code 500: cURL resource: Resource id #28; cURL error: SSL read: error:00000000:lib(0):func(0):reason(0), errno 104 (56)
  

Perhaps I should raise a photo/frontend issue? I never see this when running on a self-hosted server, so I can't look at the logs.

It would be nice if the Python library automatically retried these failed requests a couple of times. I'm not sure if we'd want to support this for all failures, or just a subset (eg. 500 errors).

If we do this, the number of retries to be configurable (or set to 0 to disable).

[jmathai]

0 by default. I like this though as most of our queue runners retry a fixed number of times against our and others' API.